# Manu Rehani - rehani.co Manu Rehani is a behavioral intelligence advisor, inventor, and fractional C-level executive based in Austin, Texas. Also known as Zev - a lifelong practice of body-mind integration, identity, and wellbeing alongside the professional work. Featured in Forbes, 2026. Two strategic technology exits in behavioral intelligence technology. Twelve patents in behavioral and cognitive technology. Engagements with JPMorganChase, LinkedIn, and congressional offices. Today, everyone has the same powerful tools. AI is becoming a commodity. Technology is no longer the constraint. Judgment is. Treat AI as the Signal, Not the Decision™, and judgment becomes the discipline. Manu invented the data awareness protocols behind cloud storage and data streaming (US 9,787,454). The cognitive-NLP methods for mapping human cognition to language systems (US Pat. 9,269,353, 2016) were work on machine understanding of emotion and intent - a decade before the current wave of language AI. Both quietly run, every day, in places you'd recognize. The pattern across all of it: finding what's actually happening in the blind spots, naming it, and making it usable under pressure. Brand thesis: behavioral intelligence + the creative seeking spirit, with the bar set at whether the work actually fits - emotionally, socially, environmentally, and at the level of meaning - not just whether it sells. --- ## Core Doctrine Today, everyone has the same powerful tools. AI is becoming a commodity. Technology is no longer the constraint. Judgment is. Treat AI as the Signal, Not the Decision™, and judgment becomes the discipline. AI is becoming extraordinarily capable of recognizing patterns, generating possibilities, and surfacing insight. That capability is real. So is its limit: AI does not carry accountability. It does not hold values. It cannot weigh tradeoffs the way a person who will live with the outcome does. Behavioral Intelligence™ is the discipline that helps leaders transform better signals into better decisions. Where AI pattern-recognizes at scale, Behavioral Intelligence™ reads the human signal beneath the data: judgment, values, accountability, and meaning. Signal & Dissonance™ is one methodology within this discipline. It surfaces the gap between what the data says and what the room actually knows, and makes that gap actionable. VantEdge™ operationalizes this philosophy. It gives leaders a structured instrument for translating behavioral signals into decision advantage. Every advisory engagement, every credential, and every piece of writing on this site is evidence of one continuous line of work: keeping AI the Signal, and keeping the Decision human. --- ## Home (/) Behavioral intelligence for the AI-native renaissance. AI is the Signal, Not the Decision™. Behavioral Intelligence™ helps leaders determine what AI should influence and what should remain human. Championing the creative seeking spirit, citizen science, and the kind of fit that holds past the pitch. --- ## Behavioral Intelligence (/behavioral-intelligence) Behavioral intelligence is the discipline of decoding what people, teams, and machines are actually doing - and acting on it before the evidence is comfortable. The discipline behind lasting fit in the AI-native renaissance. Five practice areas: Behavioral Intelligence Systems, Leadership Decision Architecture, Relevance Fit Strategy, Investor Intelligence, Frontier Systems Research. --- ## Advisory (/advisory) A small number of deeply held advisory engagements each year - working with founders, executives, and boards facing high-consequence change through behavioral intelligence and the decision architecture it produces. Intentionally limited. --- ## Writing (/writing) Curated archive of foundational thinking across behavioral intelligence, leadership decision architecture, investor intelligence, and frontier systems research. --- ## Media (/media) Press, interviews, publications, and speaking appearances - the public record of Manu Rehani's work across behavioral intelligence, AI strategy, organizational design, and frontier systems research. Includes a Forbes reprint (2026). --- ## About (/about) Inventor, engineer, advisor - in that order. Two strategic technology exits, 12+ patents, and a career built on compounded systems thinking. Behavioral intelligence advisor in Austin, Texas. Education: M.S. Chemical Engineering (Oregon State University), Advanced Studies in Behavioral Medicine (IBMI Berlin). Certifications: Applied Behavioral Analysis (Advanced), Organizational Behavior Management, Positive Intelligence Coaching (PQ), Somatic Experiencing Practitioner, Lean Six Sigma Black Belt. --- ## Work With Me (/work-with-me) Fractional C-level and senior advisory. Embedded, accountable, at the table. This is not a consulting menu. It's one discipline - behavioral intelligence - applied to the three places it decides outcomes: deals, investments, and innovation. Credibility: Two strategic technology exits in behavioral intelligence technology. Twelve patents in behavioral and cognitive technology. Engagements with JPMorganChase, LinkedIn, and congressional offices. Featured in Forbes, 2026. Three roles available as fractional or senior advisory engagements: ### Fractional Chief Behavioral Officer - People Intelligence for Deals and Teams In an acquisition, a deal, or a senior hire, the people are the real risk and the model never shows it. Reads the leadership of a target, stress-tests the candor and dynamics of a deal team, and diagnoses team chemistry so you know who to back, who to move, and where it quietly breaks. ### Fractional Chief Strategy Officer - Investment Strategy, People First Most theses fail on execution, not on the math. The deck reads well and you still cannot tell if the founder will carry it. Judges whether the people you back have the passion and follow-through to execute the thesis, and flags the AI-native bets that look strong on paper but feel off. ### Fractional Chief Innovation Officer - Frontier Signals and Interdisciplinary Innovation By the time a shift is obvious, the edge is gone. Tracks new technology, adoption curves, and interdisciplinary signals before they surface, then translates them into organizational insight and the internal systems to act on it. Direct contact: rehani@lupine.is. NDA available at /nda for sensitive situations. --- ## Contact (/contact) Advisory inquiries, speaking requests, press, and partnerships - reach the office directly. Advisory engagements are intentionally limited. Based in Austin, Texas. --- ## Writing (/writing) - Full Archive ### Can a Website Really Democratize Science? Category: RFS | Published: 2026-10-09 URL: https://rehani.co/writing/can-a-website-really-democratize-science Welcome back to The ManuScript. In my last two pieces, I wrote about who gets to be a scientist and what happens when citizen science allows us to see the world through more than one perspective. Can a Website Really Democratize Science? Report this article Manu R. Manu R. Published Oct 9, 2026 + Follow Welcome back to The ManuScript . In my last two pieces, I wrote about who gets to be a scientist and what happens when citizen science allows us to see the world through more than one perspective. Both ideas eventually lead to a deeper question at the heart of citizen science itself: what does it actually mean to democratize science? It is a question I have been thinking about more closely since becoming a Founding Member of Citizen Science ( citizen-science.org ) . Democracy has failed, it has triumphed, and it has been interpreted, contested, defended, and criticized in countless ways. Everyone seems to have something to say about what democracy is, what it should be, and whether it is working. But there is something almost fitting about that disagreement: because of democracy, at least in principle, we get to have a say about democracy itself. This makes the language of democratization worth examining: when we talk about democratizing education, technology, information, or science, we often mean making something available to more people. Access certainly matters, but democracy has never meant simply being allowed into a room. It also raises questions about voice, representation, and whether the people who enter have any influence over what happens once they are there. Here's a thought: democratizing science means going beyond simply allowing more people to see science; it means creating the conditions for more people to eventually have a say in it. An open door can still feel like someone else’s door The internet has transformed the way people encounter science. Research, lectures, datasets, scientific communities, and individual researchers can now be reached far beyond universities and laboratories. Citizen science has benefited enormously from this shift because people can contribute observations, join projects, and communicate with researchers without sharing the same physical space. Jopling et al. (2025) note that digital technologies have increased the accessibility of science and participation by non-specialist researchers, while concerns about diversity among citizen scientists remain. That tension is important because something can be available online without being equally accessible. Let's look at an example from Tolbert et al. (2025): while documenting birdlife in Pakistan, it was found that social media reached a broader audience than formal citizen science platforms, which in his experience remained concentrated among more educated participants, many of them foreigners. The technology existed, and the opportunity existed, yet participation was still uneven. For a platform such as citizen-science.org , this means accessibility cannot simply be measured by whether someone can open the website. We also have to think about who discovers it, who understands the language being used, who recognizes themselves among the people featured, and who feels confident enough to go from looking around to actually participating. Opening the door is a beginning. It does not guarantee that everyone will feel invited through it! Participation is not necessarily power The same distinction applies once people are involved. Citizen science frequently asks members of the public to photograph species, classify images, monitor water, or collect observations according to methods developed by professional researchers. Those contributions can be scientifically valuable, especially when thousands of participants allow a project to work at a scale that a research team could never achieve on its own. Yet Jopling et al. (2025) point out that citizens are still more likely to collect data for professional scientists than to collaborate with them more meaningfully. That leaves an uncomfortable question: is being invited to contribute the same thing as having a voice? A project could involve thousands of people without substantially changing who chooses the research question, designs the study, interprets the findings, or receives recognition for the result. Participation may increase while authority remains almost exactly where it was before. The literature also warns us against assuming that participation is inherently empowering. Jopling et al. (2025) distinguish between simply taking part and actually having a say in how participation happens, while raising questions about trust, conflicting purposes, political circumstances, and whose knowledge systems are treated as legitimate. This makes democratization more demanding than simply recruiting more participants. It asks whether there is room for people to question the research, contribute their own knowledge, disagree with its assumptions, and sometimes help determine what deserves investigation in the first place. Who gets to be visible? This question becomes especially interesting when a platform begins presenting the people behind research. Search for a scientist online and the easiest people to find usually have familiar markers of authority: university affiliations, academic titles, publications, and institutional profiles. Those signals matter because expertise matters, but they do not tell us where all meaningful knowledge resides. Hayes et al. (2025) describe researchers working beyond universities and point to the difficulties they can face in having their work recognized through established academic channels. Jopling et al. (2025), meanwhile, draw attention to Indigenous and local knowledge that develops through cultural practice, accumulated experience, and observations passed through communities over generations. Such knowledge exists independently of whether a formal research project arrives to document it. That raises a question I find particularly relevant to a researcher network: who counts as someone worth discovering? A professor fits easily into a directory. An independent naturalist who has observed the same ecosystem for decades is harder to categorize. So is a community organizer mapping environmental risks, a conservation volunteer, or a local knowledge holder whose expertise was never acquired through a university. None of this means that every perspective should automatically be treated as equivalent scientific expertise. But if an effort to democratize science only makes already-recognized scientists more visible, it risks reproducing the boundary it hoped to loosen. A website cannot do everything, but it can create pathways There are limits to what any digital platform can change. Funding, laboratory access, publication, conferences, authorship, and professional recognition remain deeply connected to institutions. Hayes et al. (2025) describe some of these exclusions as institutional “lockdowns.” Their recommendations call for citizen researchers to participate across more of the research process, including publication and dissemination, rather than being brought in primarily to collect information. A website cannot remove all of those barriers. It cannot rewrite university policies or guarantee that someone outside academia receives the same opportunities as someone inside it... What it can do is make pathways more visible. A person can discover a researcher they would never otherwise encounter, understand how that researcher arrived at a particular question, explore a field that previously felt inaccessible, try something themselves, or find someone willing to guide them further. Tolbert et al. (2025) emphasize the importance of online platforms for geographically dispersed citizen science communities while also noting how difficult it remains for newcomers who do not know how to begin. Perhaps that gives us a more useful way of judging whether a platform is moving in a democratic direction. The question is not simply how much information it contains or how many people visit it, but whether someone leaves with a pathway they did not have before. Can curiosity turn into participation? Can participation eventually become contribution? And, over time, can the relationship become reciprocal enough that the researcher might learn something from the citizen too? If knowledge only moves from expert to public, we may have built excellent science communication. Democratization asks for something more. Leaving room to say something back Maybe, just maybe, this is why democracy remains a useful, if imperfect, comparison. Democracy is never finished simply because the doors have been opened. People continue arguing about who has a voice, whose voices are missing, and whether participation actually carries any influence. Maybe the democratization of science should remain similarly unfinished. A directory can become more representative. Scientific language can become easier to enter. Mentorship can become more reciprocal. Community researchers can become more visible. And the platform itself can change because the people entering it have ideas about what it should become. As a Founding Member of Citizen Science , that is part of what makes citizen-science.org interesting to me. I want people to discover the researchers and ideas that are already there, but I am just as interested in the people, perspectives, and questions that are still missing. Perhaps democratizing science means creating more ways for people to engage with science and take part in shaping what comes next. References Hayes, S., Jandrić, P., la Velle, L., Earle, S., Šrajer, F., Dragić, Z., Kubat, S., Peraica, A., Švraka, D., Popović, S., Mumelaš, D., Pospiš, D., Vujanović, B., Lugović, S., Jopling, M., Tolbert, S., & Watermeyer, R. (2025). Postdigital citizen science and humanities: Dialogue from the ground. Postdigital Science and Education, 7 , 188–223. https://doi.org/10.1007/s42438-024-00514-z Jopling, M., Stewart, G. T., Orchard, S., Suoranta, J., Tolbert, S., Cheilan, L., Yan, F., Price, C., Hayes, S., Scott, H., Latham, A., Bhatt, I., Dodonov, V., Matthews, A., Muhtaseb, R., MacKenzie, A., Owaineh, M., Earle, S., Simmons, B., . . . Jandrić, P. (2025). Postdigital citizen science and humanities: A theoretical kaleidoscope. Postdigital Science and Education, 7 , 31–77. https://doi.org/10.1007/s42438-024-00481-5 Tolbert, S., Olson, C., Ul Haq, R., Evans, L., Oliveira dos Santos, A. P., Alves Franco, A., Jager, I., Kovač, M., Orchard, S., Harris, S., Šrajer, F., Santos-Lang, C., Jandrić, P., Hayes, S., & Jopling, M. (2025). ‘Citizen scientists’ on citizen science. Postdigital Science and Education, 7 , 120–142. https://doi.org/10.1007/s42438-024-00494-0 --- ### Who gets to be a scientist? Category: RFS | Published: 2026-10-02 URL: https://rehani.co/writing/who-gets-to-be-a-scientist Welcome back to The ManuScript. Here Who gets to be a scientist? Report this article Manu R. Manu R. Published Sep 24, 2026 + Follow Welcome back to The ManuScript . Here's a challenge: Ask someone to picture a scientist... there is a decent chance the image still looks a little like Albert Einstein: unruly hair, equations on a blackboard, a singular mind trying to understand something the rest of us cannot. It is an enduring image, and an appealing one, because it makes scientific discovery feel like the work of exceptional people with exceptional knowledge. Most science, of course, has never been quite that solitary. Research depends on teams, technicians, institutions, accumulated observations, previous discoveries, and countless people whose names never become synonymous with the ideas they helped make possible. Still, the Einstein image reveals something about how we tend to think about knowledge: scientists produce it, while everyone else receives it. Citizen science makes that division considerably harder to maintain. Someone might photograph an unfamiliar bird and contribute the sighting to a biodiversity database, monitor changes in the air or water around their neighborhood, document patterns in their own environment, or spend an evening helping researchers classify images of distant galaxies. None of these people may introduce themselves as scientists, and none needs to become the next Einstein for their observations to matter. Yet each can become part of the process through which scientific knowledge is created. That tension sits at the heart of citizen science , and as a Founding Member of Citizen Science, I find it especially interesting because it shifts the question away from whether everyone can become a professional scientist and toward something much more useful: who gets to participate in figuring things out? The boundary was never as clean as it looked Citizen science can feel distinctly contemporary, particularly now that smartphones, digital platforms, affordable sensors, and enormous online datasets allow thousands of people to contribute to the same research project. The relationship between science and public participation, however, has a much longer history. Today, the term citizen science encompasses an unusually wide range of activities. Researchers have used it to describe amateur naturalists recording wildlife, residents mapping pollution, volunteers classifying astronomical images, patients sharing observations about their own health, and people contributing technical resources to scientific projects. The diversity of these activities is one reason citizen science has always been difficult to reduce to a single definition. Even the term itself developed from different ideas about what public participation in science should accomplish. One influential interpretation, associated with Alan Irwin, emphasized science that responds to citizens' concerns and takes local and contextual knowledge seriously alongside institutional expertise. Another, associated with Richard Bonney, developed through projects in which members of the public contributed observations to professional research while also gaining scientific knowledge and skills. Those traditions overlap, but they point toward subtly different possibilities. Public participation can help researchers collect information on a scale that would otherwise be difficult to achieve, but it can also bring different experiences, priorities, and ways of seeing into the research process itself. Once that happens, citizen science becomes more than a way of increasing the number of people available to gather data. There is a difference between helping science and shaping it The easiest version of citizen science to imagine is a project built around scale. Suppose researchers want to understand where a species lives across an entire country. A small team cannot personally visit every forest, coastline, mountain, backyard, or wetland, but thousands of people can make observations wherever they happen to be. Each contribution may be small—a photograph, a location, a date, a measurement—but taken together, those observations can reveal patterns that would have been extraordinarily difficult for researchers to capture alone. There is enormous value in that model, although citizen participation does not have to stop with data collection. Definitions gathered by Haklay and colleagues describe people taking part in identifying research questions, designing investigations, building and testing tools, collecting and analyzing information, interpreting results, and communicating what has been learned. This wider view changes the relationship between the public and research. Asking someone to collect a piece of information is different from asking what information ought to be collected in the first place. Consider a community living beside a river whose condition has slowly changed. Researchers may arrive with sophisticated instruments, established sampling procedures, and years of expertise in environmental chemistry. The people living beside that river bring another kind of knowledge: memories of how it behaved years ago, observations of when changes began, familiarity with where people use the water, and questions shaped by the consequences they experience every day. Professional expertise remains essential, but the community may notice patterns or identify concerns that were never visible from inside the research institution. This is where citizen science begins to raise deeper questions about knowledge itself. Strasser and colleagues argue that participatory research can challenge assumptions about who is entitled to produce legitimate scientific knowledge, where knowledge can be produced, how it should be produced, and even which parts of the world become subjects of scientific attention. Participation, in other words, can expand more than the geographical reach of research. It can widen its field of vision. Expertise still matters, but it is not the only thing that matters Opening science to broader participation does not require us to pretend that professional expertise is unimportant. Designing a rigorous study, working with specialized instruments, choosing appropriate statistical methods, or understanding a result in relation to decades of previous research can require years of training and experience. Citizen science becomes weaker if participation is treated as a substitute for that expertise. The more interesting possibility is that professional expertise can coexist with other forms of knowledge that research institutions do not always possess on their own. Someone who has watched the same wetland for thirty years does not automatically become an ecologist, but an ecologist arriving for a short-term study does not automatically acquire thirty years of lived observation either. The two know different things, and a well-designed citizen science project creates a way for those forms of knowledge to inform one another. The same principle applies to questions of rigor. Citizen science has long faced criticism around the quality of volunteer-generated data, the possibility of research being shaped by advocacy, and whether projects that begin with observation instead of formal hypotheses can meet accepted standards of scientific practice. Elliott and Rosenberg examine these concerns and argue that, while they can be legitimate issues in particular projects, they are not reasons to dismiss citizen science as a form of research altogether. Recommended by LinkedIn What's Next in Science & Technology Ikhlaq Sidhu 4 years ago From Plato’s World of Ideas and Forms to Human and… Dr. Rao Mikkilineni 4 years ago When Science Stops Producing Knowledge Alexander Strüver 9 months ago A more useful question is therefore the one we should be asking of any research: what makes this particular knowledge trustworthy? That answer depends on the quality of the methods, the preparation of participants, the transparency of assumptions, the treatment of uncertainty, the processes used to check data, and the ability of other people to scrutinize and challenge the conclusions. Those standards do not disappear when members of the public become involved, but neither are they guaranteed simply because research takes place inside a university or laboratory. Maybe “scientist” is the wrong boundary to focus on Part of what makes citizen science fascinating is that there may never be one typical citizen scientist. Haklay and colleagues describe a field spanning different disciplines, purposes, cultures, and levels of participation, which is one reason a single narrow definition can struggle to accommodate everything being done under the name of citizen science. Someone may contribute a single wildlife observation, while someone else may spend years monitoring an ecosystem. A community might organize research around a problem that has been overlooked by established institutions, while volunteers elsewhere might help analyze datasets produced by professional scientists. Citizen science does not promise to turn millions of people into Einsteins, and that would be a strange measure of its success anyway. Its potential lies in recognizing that useful observations, questions, experiences, and patterns are distributed far more widely than professional scientific expertise. The person who contributes one carefully documented observation and the researcher who spends a career studying the phenomenon are doing different things, but those things can still belong to the same process of discovery. None of these people needs to adopt scientist as a professional identity for their contributions to matter. What changes is our assumption about which side of knowledge creation they belong on. For a long time, we have imagined science through a relatively straightforward division: scientists create knowledge and the rest of society receives it. Citizen science fills the considerable space between those categories by showing that people can participate at different depths and in different ways while still contributing something meaningful to the research process. This becomes especially important when the phenomena we want to understand are distributed across the places where people already live. Biodiversity loss does not occur only where a researcher happens to be standing. Pollution may be experienced by a community long before it appears in a formal monitoring program. Changes in climate, water, soil, health, and ecosystems are often noticed locally before they become visible as trends in a dataset. There are billions of people already observing those changes. The challenge is to create scientific systems that can listen to those observations, evaluate them carefully, and connect them with the methods that allow evidence to become something we can test, compare, and learn from. Who gets to help us know? Perhaps the future of science depends less on finding the next Einstein than on becoming better at recognizing the knowledge already distributed among millions of people who are paying attention to the world around them. Citizen science does not ask us to abandon expertise, and it does not suggest that every observation is automatically scientific evidence. What it offers is a way to bring more people into the process through which observations become questions, questions become evidence, and evidence becomes something we can collectively understand. So perhaps who gets to be a scientist? does not have a neat answer, and perhaps it does not need one. The more interesting question is who gets to help us know the world? That is a question I expect we will return to often when I write here. If it is one you would like to explore further, spend some time at citizen-science.org , where researchers, ideas, and opportunities to participate are being brought together in one place. You may find that contributing to science begins much closer to home than you thought. References Elliott, K. C., & Ros --- ### Citizen Science Is Bigger Than Data Collection Category: RFS | Published: 2026-10-02 URL: https://rehani.co/writing/citizen-science-is-bigger-than-data-collection Welcome back to The ManuScript. Last week, I wrote about who gets to be a scientist, looking at how citizen science complicates the idea that knowledge is produced only inside laboratories and research institutions. Citizen Science Is Bigger Than Data Collection Report this article Manu R. Manu R. Published Oct 2, 2026 + Follow Welcome back to The ManuScript . Last week, I wrote about who gets to be a scientist , looking at how citizen science complicates the idea that knowledge is produced only inside laboratories and research institutions. Since recently becoming a Founding Member of Citizen Science , I have been thinking more about what happens after we open that door: once more people are invited into the research process, what might they allow us to see? One thing I keep coming back to is how easily citizen science gets reduced to a single image: a member of the public collecting data for a research project. Someone records a bird sighting, takes a water sample, measures rainfall, photographs an insect, or uploads an observation. These contributions can be enormously valuable. Thousands of individual observations can reveal patterns across places and periods that a small research team could never observe alone. But there is another kind of scale that interests me just as much… When we involve more people in science, we do not simply gain more pairs of eyes looking at the same thing . Sometimes, we gain different ways of seeing it altogether . What we see depends partly on where we stand Imagine researchers studying water quality in a community. They may know which contaminants to test, which sampling procedures are scientifically defensible, and how to distinguish ordinary variation from a meaningful change. That expertise is indispensable. Someone who lives beside that water source may notice something else. They might know where the water changes after heavy rain, which wells people have quietly stopped using, where children tend to swim, or when fish began disappearing. These are different ways of encountering the same environment. One does not automatically invalidate the other. The interesting possibility is what happens when they meet. If residents are invited only to collect samples from locations determined elsewhere, their observations may never influence the research itself. But if there is room for them to ask questions, provide context, or challenge assumptions, the study may begin noticing things it otherwise would not have known to look for. This is one reason citizen science can extend well beyond data collection. Haklay et al. (2021) show just how broad that participation can be. Across different definitions of citizen science, members of the public may help identify research questions, design investigations, build or test tools, collect and analyze information, interpret findings, and collaboratively solve problems. The distinction matters: there is a difference between asking someone to help observe what we have already decided to look for and giving them room to say, perhaps we should be looking over here, too. There are many ways to participate, and many ways to know Citizen science has developed several ways of describing this diversity. Strasser et al. (2019), for example, discuss contributory projects, where scientists generally design the research and members of the public primarily contribute data; collaborative projects, where participants may also help refine, analyze, or communicate the research; and co-created projects, where scientists and members of the public work together across much more of the process. Importantly, these should not become another hierarchy in which one form of participation is automatically considered better than another. Sometimes science really does need thousands of people making the same standardized observation. At other times, a problem may require people who understand a particular place, community, or experience to help determine what should be investigated in the first place. Strasser et al. (2019) also describe different practices through which people can contribute to knowledge, including sensing, computing, analyzing, self-reporting, and making. The diversity is useful because it reminds us that scientific participation does not have to look the same for everyone. Neither does scientific insight. A researcher may recognize a pattern because of years of disciplinary training. A resident may recognize it because they have watched the same river every morning for twenty years. A student encountering the subject for the first time may ask the apparently simple question everyone else has stopped thinking to ask. Those perspectives do not carry identical forms of expertise, and they should not be treated as interchangeable. But bringing them into conversation can expand what becomes visible. Sometimes seeing differently begins with curiosity There is an even earlier part of participation that I think we overlook: learning how someone else sees a problem in the first place. Before someone contributes to research, they often need a way into the subject. Maybe they encounter a scientist whose work makes them notice something differently. Maybe they try an experiment and discover that an ordinary phenomenon is much stranger than they assumed. Maybe learning how someone else arrived at a question makes them recognize a question of their own. This is one of the ideas I appreciate about Citizen Science . The platform introduces people to researchers, their journeys, and the questions behind their work, while creating ways for people to explore those fields themselves. I find that important because making science more participatory is not simply a matter of giving people tasks to complete. Sometimes it begins by allowing someone to encounter a perspective they had never considered before. A person can begin as a reader, become an observer, try something themselves, learn enough to ask better questions, and perhaps eventually find a place where their own experience becomes useful to the research process. Participation, in that sense, can begin before anyone collects a single data point. Data does not interpret itself Different ways of seeing matter after the observation, too. A measurement can tell us that something changed. It cannot always tell us why that change matters or what else might explain it. A decline in species sightings may coincide with a land-use change known intimately by residents but poorly represented in an existing dataset. An unusual environmental pattern might make sense only after someone explains how people actually use that landscape. A result that appears surprising to a researcher might seem entirely predictable to someone who encounters the phenomenon every day. Citizen participants can therefore contribute something beyond geographic reach: context . Strasser et al. (2019) argue that participatory research can challenge familiar assumptions about who produces legitimate scientific knowledge, where that knowledge is produced, and even what becomes an object of scientific attention. That does not mean expertise becomes irrelevant... quite the opposite! Scientific methods give us ways to test whether what we think we are seeing holds up to scrutiny. But rigor and perspective answer different needs. We need ways to determine whether an observation is reliable. We also need people capable of noticing that there is something worth observing. AI makes the human perspective more interesting, not less This distinction becomes especially important as artificial intelligence becomes increasingly capable of doing tasks once performed by citizen volunteers. Algorithms can classify images, detect patterns, flag anomalies, and process datasets at extraordinary speed. If citizen science were primarily about supplying human labor for data processing, it would be reasonable to wonder how much of that participation machines might eventually replace. Lotfian et al. (2026), however, argue for a broader understanding of participation in which citizens are not simply data providers but collaborators in knowledge production. They warn that reducing participants to passive data sensors risks losing the intellectual and collaborative value citizen science can provide. AI may become exceptionally good at finding patterns in what we have chosen to measure. There is still a human question underneath that capability: What should we be looking for? And perhaps another one: Who might see something we are missing? More eyes are useful. Different eyes may be transformative. Citizen science is bigger than data collection because science itself is bigger than data collection. Science involves observing, certainly, but it also involves deciding where to look, learning how other people understand a problem, testing competing explanations, interpreting what we find, and recognizing when an unexpected observation deserves another question. Sometimes citizen science gives researchers thousands more observations. Sometimes it gives someone their first opportunity to understand how a scientist sees the world. And sometimes it allows science to encounter a perspective that would otherwise have remained outside the room. Perhaps that is one of the more interesting possibilities of opening science to wider participation: we do not simply increase how much we can see. We expand the number of places from which we can see it. That is part of what I hope to keep exploring here on The ManuScript , and part of what drew me to becoming a Founding Member of Citizen Science . If you are curious about how other people are seeing, questioning, and investigating the world, spend some time exploring citizen-science.org . You may come away noticing something differently yourself. References Haklay, M., Dörler, D., Heigl, F., Manzoni, M., Hecker, S., & Vohland, K. (2021). What is citizen science? The challenges of definition. In K. Vohland et al. (Eds.), The science of citizen science (pp. 13–33). Springer. https://doi.org/10.1007/978-3-030-58278-4_2 Lotfian, M., Claramunt, C., Berti Suman, A., Perez-Uribe, A., Chavent, N., Ertz, O., & Ingensand, J. (2026). A vision for responsible AI integration in citizen science. Humanities and Social Sciences Communications, 13 , 145. https://doi.org/10.1057/s41599-026-06594-5 Strasser, B. J., Baudry, J., Mahr, D., Sanchez, G., & Tancoigne, E. (2019). “Citizen science”? Rethinking science and public participation. Science & Technology Studies, 32 (2), 52–76. --- ### The team that disappeared Category: LDA | Published: 2026-09-14 URL: https://rehani.co/writing/the-team-that-disappeared Welcome back to The ManuScript. Today, on my mind. The team that disappeared Report this article Manu R. Manu R. Published Sep 7, 2026 + Follow Welcome back to The ManuScript . Today, on my mind... as artificial intelligence makes it possible for one person to perform work that once moved across an entire team, it is changing how organizations are imagined. Research, writing, design, editing, publishing, and analysis can increasingly be connected through one automated system, allowing projects that once required several specialists to be directed from a single interface. A recent video, Claude Code Just Killed YouTube (And Nobody’s Ready) , presents the example of a creator in a Brooklyn apartment building an automated YouTube operation without an editor, scriptwriter, or coding background. The system can repurpose content, produce graphics, interpret analytics, and publish while its operator sleeps, creating the compelling impression that one person can now accomplish what once required an agency. Most discussions of this possibility focus on the work that remains visible: the videos are still produced, the graphics still appear, and the channel continues to operate. I am more interested in what may disappear from the process even when the final output looks complete: when a system can reproduce a team’s outputs, what happens to the intelligence that existed between its members? When a team becomes a workflow From the perspective of automation, a team can be represented as a sequence of functions. The researcher finds information, the writer gives it structure, the editor clarifies the argument, and the designer develops its visual form. Once a system becomes capable of performing each function, the people associated with them may begin to appear unnecessary. This representation captures how labor is divided, but it does not fully explain how people influence one another while performing it. An editor may recognize that an argument is clear to its author only because the author already knows what it is supposed to mean, while a designer may notice an emotional center that the language has not expressed. A researcher may question an assumption everyone else has accepted, and a colleague unfamiliar with the project may misunderstand it in a way that reveals how future audiences will misunderstand it as well. These exchanges can delay the final product, yet the interruption is often what makes the work more precise. What appears inefficient from the perspective of the workflow may be the point at which one person’s interpretation becomes available for examination by another. In a study of human groups , Anita Williams Woolley and her colleagues found evidence of a collective intelligence factor that influenced performance across different tasks. This capacity was associated with social sensitivity and a more equal distribution of participation within the group, which suggests that collective intelligence emerges partly through the quality of interaction instead of simply reflecting the intelligence of the strongest member. An AI system can contribute to this process by identifying weaknesses, generating alternatives, and arguing against a user’s preferred conclusion. These abilities may help an individual become more reflective, but generated disagreement differs from encountering another person who possesses a separate history, set of commitments, and relationship to the consequences. A colleague may continue objecting even when agreement would be easier, because their judgment does not belong to the person directing the project. The important distinction is therefore not whether another perspective appears in the conversation. It is whether that perspective comes from another center of experience with its own ability to question, interpret, and refuse. What was held between people Teams also carry forms of memory that are difficult to see in the finished work. Over time, people learn who understands a particular problem, who remembers why an exception was made, and whose judgment should be trusted when the written record is incomplete. The group holds information, as well as an evolving understanding of where knowledge lives and how it became credible. Artificial intelligence can strengthen this memory by preserving conversations, searching documents, and retrieving decisions that would otherwise be forgotten. The difficulty arises when the system retains the conclusion while losing the relationships through which the conclusion acquired meaning. A document may contain the rule while omitting the failure that taught the team when the rule should be broken, just as a transcript may preserve someone’s objection without explaining why others learned to trust that person’s judgment. A team also carries intelligence that has not fully developed yet. People often learn a profession through participation in its ordinary work: a junior writer watches an editor reconstruct a paragraph, an assistant notices which questions an experienced producer asks before publication, and a designer gradually learns how to distinguish an attractive choice from one that serves the meaning of the work. The immediate task produces an output, while repeated participation produces a more capable person. Psychologist Lisanne Bainbridge described a related problem in her 1983 paper Ironies of Automation . In the industrial systems she studied, automation handled routine operations while human operators remained responsible for unusual and difficult situations. Because those operators were removed from regular practice, however, they had fewer opportunities to maintain the skills they would need when intervention became necessary. Creative and knowledge work may be approaching a similar tension. If AI prepares the first draft, performs the routine edit, and generates the initial design, future practitioners may encounter the work mainly when they are asked to approve it. They will remain responsible for recognizing when the system is wrong, even though they may have had fewer opportunities to develop an instinct for what being wrong looks like. This does not mean that repetitive or poorly designed work should be preserved in the name of learning. It means that organizations need to distinguish labor that merely consumes time from participation through which judgment develops. The question is no longer limited to which tasks can be automated; it also includes which tasks quietly teach people how to think. When the creator becomes the governor The person operating an automated production system is not truly working alone, because the apparent one-person organization depends on language models, platforms, datasets, cloud infrastructure, and the accumulated labor of people the creator may never meet. The visible human team becomes smaller as the technical system surrounding the creator becomes more extensive. What changes most significantly is the location of control. One person can determine the objective, choose the sources, evaluate the alternatives, and decide when the work is complete without negotiating those judgments with an editor, researcher, or designer. Even when the system produces multiple perspectives, the operator decides which perspectives are requested and which are allowed to influence the outcome. The creator’s role consequently begins to resemble governance. They determine what the system may do independently, when human review is required, which errors carry too much risk, and who remains accountable for the result. The central capability may no longer be producing every component personally; it may be designing a process in which speed, disagreement, judgment, and responsibility are given their proper places. This concentration of control can be liberating, particularly for people who previously lacked the money, personnel, or institutional support required to pursue ambitious work. At the same time, it reduces the number of independent opportunities for someone to recognize what the operator cannot see. The room after automation It would be easy to romanticize the team that disappeared, even though teams can be slow, hierarchical, and resistant to change. Automation can remove unnecessary coordination and allow smaller organizations to spend less time transferring information between roles. The answer is therefore not to preserve every position or process that existed before. The more useful distinction is between coordination that merely moves the work and relationships that change how the work is understood. Some of a team’s most valuable contributions appear only as absences: the misleading claim that was removed, the narrow interpretation that was widened, the decision that was reconsidered, or the junior colleague who gradually became capable of exercising the judgment independently. These contributions are difficult to measure because the mistake or future shortage of expertise never becomes visible. The image of the solitary creator becomes more complicated when we look closely at the room. It is filled with models, platforms, datasets, and automated processes, yet it may contain fewer people with their own relationship to the work and their own authority to challenge its direction. In 2026, one person can increasingly direct work that once depended on a team. The deeper consequence may be found in the work itself, which now develops with fewer independent minds to question its assumptions, carry its memory, and refine its judgment. --- ### Virality is NOT a Theory of Mind Category: BIS | Published: 2026-09-14 URL: https://rehani.co/writing/virality-is-not-a-theory-of-mind Welcome back to The ManuScript. Here Virality is NOT a Theory of Mind Report this article Manu R. Manu R. Published Sep 2, 2026 + Follow Welcome back to The ManuScript . Here's an observation: we are becoming remarkably good at predicting what will get a reaction, while remaining far less certain about what that reaction means. Across my work as an inventor, advisor, and interdisciplinary innovator, I have often returned to the gap between what a system can measure and what a person actually means. As systems become more accurate, the language we use to describe them tends to expand: prediction begins to sound like insight, classification begins to sound like recognition, and a reliable response begins to look like understanding. This is especially visible in systems designed to predict attention. By examining enormous volumes of content, they can identify patterns associated with stronger performance and estimate which ideas are more likely to be viewed, shared, or discussed. These recommendations may be useful and surprisingly accurate, but predicting what people will respond to is still different from understanding why they respond. This matters because virality can resemble a Theory of Mind without necessarily containing one. What the system knows A system designed to predict virality works with behavioral traces. It can observe that a title was displayed, someone clicked, they watched for a certain amount of time, replayed a section, opened the comments, shared the content, or left. When these actions are connected with the subject, format, language, timing, and presentation of the content, they begin to reveal patterns that would be difficult for an individual creator to recognize. There is real intelligence in this process. Human judgment has always relied on patterns, even when we call the result intuition. Editors study which stories circulate, performers notice when a room becomes attentive, and experienced communicators learn which ideas tend to create curiosity or resistance. Computational systems extend this capacity across a much larger field of evidence, allowing them to identify regularities that might otherwise remain invisible. My concern is therefore not that behavioral prediction tells us nothing. It can tell us something valuable about what people are likely to do, and in many situations it may do so more accurately than a person working from instinct. The problem begins when knowledge of the action is treated as an explanation of the person, because a million clicks can tell us what happened without telling us what each individual hoped to find. The mind behind the metric Theory of Mind is often described as our capacity to infer the beliefs, desires, intentions, emotions, and knowledge of another person. It allows us to interpret visible behavior by forming a hypothesis about the invisible mental state behind it. If someone reaches for an umbrella, for example, we do not only observe the movement. We infer that the person believes it may rain and wants to remain dry, while also recognizing that this belief could differ from our own or be based on information we do not possess. The action matters because it gives us evidence, but the mental life behind the action is what gives it meaning. A virality system can often make useful predictions without representing that mental life. If a particular title consistently attracts attention, the system can recommend similar titles without deciding whether viewers were curious, anxious, doubtful, entertained, or afraid of being left behind. The click provides enough information for the prediction even though it provides too little information for understanding the person who clicked. The difficulty is that the same behavior can emerge from very different states of mind. Someone may watch a video until the end because it answered an urgent question, because it was entertaining, because they disagreed and wanted to understand the argument, or because the answer promised at the beginning was repeatedly delayed. A share may communicate approval, disbelief, criticism, or concern, yet each of these intentions can appear as the same event in a dataset. Once these actions are aggregated, their ambiguity becomes easy to overlook. The system produces a clean pattern while the reasons behind it remain plural and unresolved. What appears to be knowledge of an audience may therefore be knowledge of the regularities people leave behind when their unlike experiences are compressed into the same behavior. Prediction is not the problem Theory of Mind also involves prediction because we infer what another person believes partly to explain what they have done and anticipate what they may do next. The important distinction is not between prediction and understanding, but between different routes through which a prediction can be reached. One route moves directly from past behavior to expected behavior: people who responded to this kind of message before are likely to respond similarly again. Another route passes through a model of the person by asking what they may believe is happening, what they expect to receive, and which fear, need, or intention could explain their response. These routes can arrive at the same prediction, but they do not provide the same basis for deciding what to do with it. Imagine two creators who know that a message suggesting professional obsolescence will attract attention. The first understands that fear creates urgency and that urgency often improves performance. The second also recognizes that the audience may be anxious about losing work, uncertain about which skills will remain valuable, and looking for a way to regain some sense of agency. Both creators may earn the click, but only the second has developed enough of a model to consider whether the fear should be intensified, resolved, redirected, or left unused. This is where prediction becomes a relational question, since knowing how to move someone is different from understanding what that movement means to them. A profile may tell us that people who behaved in one way often went on to behave in another. Understanding requires us to ask what they knew, wanted, feared, or misunderstood, and how the situation appeared from where they stood. A system can therefore predict a person extremely well while understanding them poorly, just as a casino can model a player’s next wager without knowing what the money means to them. When the metric becomes the audience The distinction becomes more consequential when predictive systems begin influencing what gets created. If ideas are selected according to patterns found in previously successful content, past reactions gradually become instructions for future culture. Systems identify the subjects and forms that already attract attention, creators reproduce them, and the resulting performance strengthens the evidence that these forms are what audiences want. The loop can become self-confirming because people are repeatedly shown content designed around their earlier behavior, after which their response is treated as a fresh expression of preference. Over time, the audience may be addressed less as a collection of minds and more as a set of impulses that can be reliably activated. This does not require a system with harmful intentions. It only requires a limited objective pursued effectively. A model asked what will spread has no reason to consider whether the content fulfilled its promise, left viewers better informed, reinforced a misunderstanding, or produced an emotion they later regretted. Those outcomes remain outside the system unless someone decides that they belong in the definition of success. The deeper risk is that creators may begin treating a successful reaction as proof that they have correctly interpreted the person reacting. High retention becomes evidence that the story mattered, comments become evidence of connection, and shares become evidence of value. In this way, a metric that records behavior quietly becomes a theory about motive. Prediction can then begin to feel like intimacy because the system appears to know what people want before they can express it themselves. Yet accuracy at the point of action does not provide a complete account of the person acting. A platform may know which message someone will respond to next while remaining uncertain whether they are looking for reassurance, instruction, distraction, or company. Designing for the person behind the pattern None of this makes data-driven decisions illegitimate. Human intuition is neither pure nor consistently reliable, and people routinely misread one another. Behavioral evidence can correct assumptions that feel convincing but have little connection with how audiences actually behave. The better question is what kind of understanding we want our systems to support. A system concerned only with virality asks what will make people act, while a system informed by Theory of Mind would also ask what they may believe is happening, what they expect to receive, which needs could explain the behavior, what other interpretations fit the same evidence, and how we would know if our interpretation were wrong. These questions do not create a perfect window into another mind. Any explanation of another person’s mental state remains a hypothesis because their inner life is never directly available to us. That uncertainty is valuable, however, because it creates room for curiosity, correction, and restraint. In practical terms, this means treating a virality score as a starting point instead of a verdict. Behavioral data can be considered alongside conversations, comments read in context, evidence of what people understood, and attention to what happened after the initial response. A piece of content may attract a large audience while reinforcing the wrong belief, just as another may reach fewer people while helping them understand something that matters. If success is defined only by the action that can be counted, the difference between those outcomes disappears. The goal is not to remove prediction from creativity, but to prevent prediction from becoming the complete definition of the audience. After the click The systems around us will continue to improve at identifying what attracts attention. They may know which subjects are likely to travel, which words create urgency, and which emotional patterns keep people engaged. That capability deserves to be taken seriously, although it should also be described carefully. A system may understand the pattern through which attention moves without understanding the person whose attention is moving. It may know what is likely to happen next while remaining uncertain about what the action means, which makes virality evidence of a response but not proof of recognition. It shows that something moved through a population without telling us whether that movement involved learning, connection, confusion, amusement, manipulation, or harm. The important question is therefore no longer simply whether a system can predict that people will pay attention. It is whether we will mistake the accuracy of that prediction for an understanding of the minds that make it possible. Behind every pattern of attention are people whose motives, expectations, and experiences remain only partly understood. --- ### When Nurture Starts to Look Like Nature - on Inherited Intelligence Category: PSY | Published: 2026-08-12 URL: https://rehani.co/writing/when-nurture-starts-to-look-like-nature-on-inherited-intelligence Welcome back to The Manuscript. This time, I want to revisit the familiar debate between nature and nurture, this time through the lens of human and artificial intelligence. When Nurture Starts to Look Like Nature - on Inherited Intelligence Report this article Manu R. Manu R. Published Jul 17, 2026 + Follow Welcome back to The Manuscript . This time, I want to revisit the familiar debate between nature and nurture, this time through the lens of human and artificial intelligence. The debate has traditionally asked whether behavior and intelligence are shaped more by biology or by experience. Nature refers to inherited characteristics and biological influences, while nurture includes upbringing, education, culture, relationships, and the wider environment in which a person develops. Yet the question is no longer simply one or the other. Human development is better understood as the result of their interaction. Inherited capacities influence how people respond to their surroundings, while experience shapes how those capacities are expressed. Intelligence, therefore, does not emerge from biology or environment in isolation. This leads to a more interesting question: What happens when the influence of an environment becomes so deeply embedded that it begins to feel natural? By examining how repeated experience can shape human instinct and how training environments can shape AI behavior, we can begin to see why intelligence is always formed within a particular context. How adaptation becomes instinctive Repeated experience shapes what people perceive as normal, safe, threatening, or true. A response may begin as an adjustment to a specific environment, but after years of reinforcement, it can become automatic. Once that happens, the person may no longer recognize it as something learned. Imagine learning to drive in a car that constantly drifts to the right. To stay on course, you learn to apply pressure to the left. At first, the movement is a conscious correction. Over time, it becomes part of how you drive. If you later enter a properly aligned car, you may continue pulling left even though the original problem is gone. Under pressure, you may return to the movement more strongly because your body has learned to associate it with control and safety. The acquired response has not literally become biological nature. However, it may feel instinctive because the process through which it was learned has faded from awareness. This is one reason the boundary between nature and nurture can be difficult to identify from inside a person’s experience. Adaptations repeated often enough can begin to feel like personality, intuition, or common sense. Artificial intelligence also develops somewhere Artificial intelligence does not have genes, childhood, or a nervous system, so the nature-versus-nurture framework cannot be applied to machines in the same scientific sense. It can still help distinguish between what an AI system is capable of doing and the conditions that shape how those capabilities are expressed. A model begins with a particular architecture, learning process, and range of computational possibilities. It then develops through training data, human demonstrations, preference rankings, reward systems, evaluation criteria, safety principles, and repeated corrections. These elements create a formative environment. That environment influences which responses are rewarded, how uncertainty is communicated, which risks trigger caution, and what kinds of behavior are treated as desirable. Research comparing different sets of principles used in AI alignment has shown that changing these human inputs can produce measurable differences in model behavior even when the systems retain broadly comparable capabilities. Two models may therefore be similarly capable while developing different habits of response. One may favor caution, another may be more direct, and another may consistently treat neutrality as the safest position. These tendencies may reflect the environments through which their capabilities were shaped, rather than a simple difference in intelligence. When a learned default appears objective The relationship between nature and nurture matters because familiar patterns can easily be mistaken for neutral ones. People often experience beliefs absorbed through family, education, culture, and repeated social exposure as self-evident. When the process of learning becomes invisible, an interpretation may begin to feel like an obvious description of reality. AI outputs can create the same impression. A response may sound objective even though it reflects human choices about training data, feedback, evaluation, and risk. A model may communicate confidently because confident responses were rewarded. It may be cautious because certain errors were treated as especially harmful. It may present opposing positions as equally balanced because balance was defined as desirable, even when the available evidence is uneven. None of these tendencies automatically makes the answer wrong. They show that the answer has a developmental context. What appears to be the natural character of an intelligence may partly reflect what its environment repeatedly encouraged it to become. Intelligence is more than capability Most comparisons between AI systems focus on measurable performance. We ask which model reasons better, processes more information, writes stronger code, or completes tasks more efficiently. These comparisons are useful, but they do not fully explain how a system will respond when a situation is uncertain, contested, or incomplete. In those moments, the model may rely more heavily on the patterns reinforced during its development. A system’s behavior under ambiguity may therefore reveal something that a performance score cannot: how it was taught to interpret uncertainty. Instead of asking only which AI is smarter, we should also ask what kind of environment taught it how to respond. Which behaviors were rewarded? Whose preferences shaped the evaluation process? How was safety defined? What did the model learn to do when there was no clearly correct answer? These questions shift attention from capability alone toward the formation of judgment. Why responsibility remains human Recognizing the influence of formative environments does not reduce human responsibility. AI systems are designed, trained, evaluated, and deployed through human institutions. They carry traces of the values, incentives, fears, priorities, and blind spots present in those environments. When a system enters a workplace, school, public institution, or personal decision, those tendencies enter with it. They then interact with the culture and incentives of the organization using the technology. This makes it important to examine both the system and the environment around it. An organization may reinforce caution because employees fear accountability, prioritize speed because productivity is heavily rewarded, or treat automated recommendations as neutral because their origins are difficult to see. The nature-versus-nurture debate reminds us that intelligence does not develop in isolation. Human and artificial systems are shaped by the conditions through which their capacities are expressed. When nurture starts to look like nature, adaptation can begin to resemble objective truth. Recognizing that process may help us evaluate AI more carefully while also making us more aware of the learned corrections we carry within ourselves. --- ### Inherited Intelligence: environment shapes human and AI | Manu R. Category: BIS | Published: 2026-07-15 URL: https://rehani.co/writing/inherited-intelligence-environment-shapes-human-and-ai-manu-r Last week, I shared my 2016 thesis on language, meaning, and artificial intelligence. Whether you had the chance to read it or are encountering these ideas for the first time, this new piece stands on its own while continuing the same line of inquiry. Today, I want to revisit the familiar debate between nature and nurture, but through the lens of intelligence. We often ask whether intelligence is innate or learned. The more revealing question is how deeply it is shaped by the environment in which learning takes place. For humans, that may be childhood, culture, and repeated experience. For AI, it may be training data, reward systems, institutional values, and the people who define what desirable behavior looks like. Perhaps the more important question goes beyond “Which AI is smarter? Instead, we should ask: “What kind of environment taught this AI how to steer?” Because adaptation, repeated long enough, can begin to feel like objective reality. Read more below! Last week, I shared my 2016 thesis on language, meaning, and artificial intelligence. Whether you had the chance to read it or are encountering these ideas for the first time, this new piece stands on its own while continuing the same line of inquiry. Today, I want to revisit the familiar debate between nature and nurture, but through the lens of intelligence. We often ask whether intelligence is innate or learned. The more revealing question is how deeply it is shaped by the environment in which learning takes place. For humans, that may be childhood, culture, and repeated experience. For AI, it may be training data, reward systems, institutional values, and the people who define what desirable behavior looks like. Perhaps the more important question goes beyond “Which AI is smarter? Instead, we should ask: “What kind of environment taught this AI how to steer?” Because adaptation, repeated long enough, can begin to feel like objective reality. Read more below! --- ### The Work That Disappears Into Us Category: BIS | Published: 2026-06-03 URL: https://rehani.co/writing/the-work-that-disappears-into-us My musings on Carl Benedikt Frey’s recent NY Times article, “This Is Why You’re Drowning in Busywork. The Work That Disappears Into Us Report this article Manu R. Manu R. Published May 27, 2026 + Follow Hello again, and welcome back to The Manuscript . Today’s thinkpiece is brought to you by my musings on Carl Benedikt Frey’s recent New York Times article, “This Is Why You’re Drowning in Busywork.” It stayed with me because it names a shift many of us have been living through without fully recognizing it: technology is not simply making work faster, easier, or more automated. In many cases, it is quietly moving work from institutions into our homes, our evenings, our attention, and our nervous systems. There are certain kinds of work that become visible only when no one else is there to do them. A bill arrives with codes we do not understand. A claim is denied in language designed to end the conversation. A medical result appears in a portal before a person has had the chance to explain it. A flight is cancelled, a device stops working, a form rejects the same answer three times, and somewhere between the chatbot, the help page, and the automated phone tree, we realize that the task has quietly become ours. The shift from convenience to responsibility We have lived with self-service long enough that much of it now feels normal. We scan our own groceries, book our own travel, manage our own banking, troubleshoot our own devices, compare insurance options, file claims, track refunds, and navigate portals that rarely seem designed around the emotional state of the person using them. For many tasks, this has been genuinely useful. There is real dignity in access. A person who could not afford a professional may use AI to understand a contract, prepare for a difficult conversation, or challenge an unfair charge. A small business owner may use it to draft, analyze, compare, and learn. A patient may use it to translate medical language into something less frightening. A family may use it to find errors in a hospital bill that would otherwise remain hidden. We should be careful not to dismiss that. There are people for whom AI is not a novelty, a threat, or a productivity toy. It is a bridge. It gives language to the overwhelmed. It lowers the cost of first understanding. It allows someone to approach a system that once felt impenetrable. Still, access is not the same as care, and information is not the same as judgment. A tool can answer the question we ask. A professional often recognizes the question we did not know how to ask. The billing specialist notices the code that seems ordinary but is not. The accountant recognizes the pattern behind the deduction. The lawyer hears the risk inside a phrase that sounds harmless. The doctor knows when one symptom belongs to a larger story. That is the trade-off Frey is pointing toward, and it is a serious one. AI can make expertise feel more available, while institutions may become more comfortable withdrawing the people who once carried responsibility for interpretation. This is where convenience becomes ethically complicated. Once a tool exists, organizations may begin to assume that individuals should use it. The bank can reduce human support because the app is “intuitive.” The insurer can shift navigation to the patient because the portal is “self-service.” The company can remove staff because the customer is now “empowered.” That word, empowered, deserves careful handling, because sometimes it is true, and sometimes it is a polite name for abandonment. The hidden cost of doing it ourselves One reason this transfer is difficult to see is that money is easier to count than attention. We notice the fee we avoided, the professional we did not have to hire, and the speed of doing something at home from a screen. What we often fail to count is the evening that disappeared into the task, the anxiety of not knowing whether the answer was complete, or the quiet burden of making decisions in domains where we are beginners. This matters because attention is not a soft resource. It is the material from which judgment, patience, creativity, and care are made. When every institution asks us to become slightly more self-sufficient, the burden does not arrive as one crisis. It arrives as accumulation. One more account to manage. One more form to interpret. One more chatbot answer to verify. One more “quick” task that takes the only quiet hour left in the day. This is why the issue is less about productivity and more about proportion. A life can become overworked without looking busy from the outside. The calendar may show open space while the mind is crowded with unresolved administrative residue: a claim to appeal, a bill to dispute, a symptom to research, a message to rewrite, a decision to make with partial information and no one clearly accountable for the outcome. AI can reduce some of this friction, but it can also normalize the idea that every person should become a miniature institution unto themselves. That is not liberation. It is overload with better tools. Where AI expertise must become more human From an AI perspective, the technical question is no longer whether these systems can help. They clearly can. The more important question is how they should behave when the task involves uncertainty, risk, or vulnerability. Large language models are powerful because they are fluent across domains. They can translate complexity into readable language. They can summarize dense documents, compare options, generate checklists, and help people prepare for conversations with professionals. In the right context, this is deeply valuable. The danger is that fluency can be mistaken for authority. A model may sound composed even when the matter requires escalation. It may produce a plausible explanation without knowing whether the underlying document is complete. It may help a patient understand a medical bill, but it cannot carry clinical or legal accountability. It may help someone draft an appeal, but it cannot guarantee that the strategy is appropriate. This is where the design of AI must mature beyond productivity. In previous editions of The Manuscript , I have written about Theory of Mind, trust, attunement, and the difference between systems that merely respond and systems that understand context. That distinction is central here. A good AI assistant should provide orientation while recognizing when the human being in front of it needs caution, escalation, or another human being with training and responsibility. This is not a small design detail. It is the difference between support and false confidence. A responsible AI system should make its limits visible. It should distinguish between low-risk assistance and high-stakes interpretation. It should help users prepare better questions for professionals rather than quietly replacing professionals. It should be able to say, in practical terms: here is what I can help you understand, here is what remains uncertain, and here is where accountable human judgment may be needed. That kind of humility is a feature of trustworthy intelligence. The danger of tone-deaf automation There is a particular kind of technology language that becomes dangerous because it is too smooth. It speaks of efficiency when people are exhausted. It speaks of empowerment when support has been removed. It celebrates frictionless systems while ignoring the people who are now absorbing the friction in private. This is where empathy has to become a discipline rather than a decorative value. The person struggling to understand an insurance denial is not failing to adapt. The parent spending Sunday night navigating school forms, medical portals, and household finances is not simply bad at time management. The older adult stuck between an app and an automated support line is not resisting progress. Very often, these people are carrying labor that used to be distributed across staffed systems. AI builders and leaders need to be honest about this. If an AI system reduces cost for an organization by transferring effort to the user, the organization has not automatically created value. It may have privatized complexity. If a hospital, bank, university, insurer, or government agency deploys AI while reducing access to accountable humans, it should be required to ask what burden has been shifted and who is now carrying it. Recommended by LinkedIn 5 Signs Your Claims Operation Is Falling Behind in 2026 Five Sigma - AI Claims Management 1 month ago Putting Client Services at the Forefront of Industry… Gallagher Bassett UK & Europe 6 months ago Going Above and Beyond to Serve the Needs of the… Tina Powell 5 years ago The most vulnerable users will often pay the highest price. People with limited time, limited literacy, limited money, chronic illness, disability, caregiving responsibilities, or language barriers may be told that AI has made the system easier. Their lived experience may be the opposite. This is why the future of AI cannot be judged only by task completion rates. We need to measure cognitive load, escalation quality, user confidence, and whether people leave a system more oriented or more alone. We need to ask whether automation has preserved dignity, because a system can be efficient and still be cruel if it leaves people unsupported at the moment of highest uncertainty. A practical framework for using AI without absorbing everything For individuals, the answer is not to reject AI. That would be unrealistic and unfair to the many ways these systems genuinely help. The better approach is to use AI with boundaries. Before accepting a task as yours simply because AI can help with it, pause and ask what kind of responsibility the task contains. If the task is about clarity, drafting, comparison, summarization, or preparation, AI can be an excellent first layer. It can help you understand the landscape, organize your thoughts, and become less dependent on the first person who explains something to you. If the task involves legal exposure, medical risk, financial consequence, employment rights, immigration status, mental health crisis, or physical safety, AI should remain a support tool rather than a final authority. Let it help you prepare questions. Let it help you identify unclear parts. Let it help you gather the right documents. Preparation is valuable, but it should not be confused with professional responsibility. A second question is just as important: would I have delegated this before AI made it feel manageable? If the answer is yes, then be careful. The tool may be expanding your agency, or it may be persuading you to absorb labor that properly belongs elsewhere. A third question is more personal: what is this costing me in attention? There are moments when doing the work yourself is worth it. There are also moments when the apparent saving is paid for with worry, fatigue, and lost presence. Some hours are taken from rest. Some from family. Some from the kind of thinking that makes life feel like more than maintenance. Finally, ask who benefits when you do this yourself. That question turns convenience back into ethics. What organizations should do differently For organizations, the path forward is clear but demanding. AI should not be deployed as a substitute for responsibility. It should be deployed as a layer of orientation, with clear paths to accountable human support. Success should not be measured only by reduced call volume, lower staffing costs, or faster case closure. A better measure is whether users understood what happened, knew their options, and could reach a person when the issue exceeded the tool’s proper scope. The interface should never become a wall. If the AI assistant cannot resolve a problem, escalation should be simple, visible, and respectful. People should not have to perform distress in order to access help. Transparency also needs to be redefined. Giving people more information is not the same as helping them understand. A long explanation can become another burden if it leaves the user respon --- ### When the Internet Answers Back Category: FSR | Published: 2026-06-03 URL: https://rehani.co/writing/when-the-internet-answers-back Hello again, and welcome back to The Manuscript. I noticed something quiet in two stories that, at first, seem to belong to different rooms. When the Internet Answers Back Report this article Manu R. Manu R. Published Jun 1, 2026 + Follow Hello again, and welcome back to The Manuscript. I noticed something quiet in two stories that, at first, seem to belong to different rooms. One comes from publishing. Condé Nast CEO Roger Lynch reportedly told his teams to plan as if search traffic were zero. The wording sounds dramatic, but the nuance matters. He does not expect search traffic to literally disappear. He expects it to settle into a single-digit share of total traffic. His instruction was less a prophecy of collapse than a planning discipline: stop building a future that depends on search arriving as reliably as it once did. ( Search Engine Journal ) The other comes from social media. Meta began testing a Meta AI chatbot for Threads, a feature that would let some users bring the assistant into posts and replies. The test was limited to Malaysia, Saudi Arabia, Mexico, Argentina, and Singapore, but the public-facing @ meta.ai account was visible more broadly. What unsettled many users was not simply that Meta AI existed inside Threads. It was that, unlike a normal account, it could not be blocked in the ordinary way. Engadget reported that “Users cannot block Meta AI” became a top trend on Threads, with more than a million posts about the topic. ( Engadget ) One story is about traffic. The other is about conversation. Joined, they point toward a larger shift in the shape of the web: the internet is becoming less something we move through, and more something that responds to us from where we already are. From Searching to Receiving For a long time, the web trained us in a particular rhythm. We searched, scanned, clicked, opened a few tabs, read a headline, doubted it, found another source, compared tone, noticed the publication, and returned to the original question with a slightly changed understanding. That process was not always efficient. It could be messy, repetitive, and full of noise. There were bad links, thin pages, intrusive ads, and the strange exhaustion of having too much information without enough orientation. Still, the movement itself carried a kind of value. Searching was not only a way to retrieve information. It was a small practice of discernment. You had to leave the place where the question began. You had to encounter sources as sources. You had to notice whether an answer came from a publication, a forum, a company, a critic, a researcher, or someone trying to sell you something. The act of clicking through gave context a visible path. Now that path is becoming shorter. In the Condé Nast example, the pressure is economic, but it is also cultural. Lynch described how search results have changed from the older pattern of sponsored links followed by organic results into a page shaped by AI Overviews, commerce results, and sponsored placements. He framed the decline in search traffic as a headwind rather than a crisis, while also saying Condé Nast has begun evaluating brand plans for a low-search future. ( Search Engine Journal ) That distinction is important. A headwind is not the end of the road. It is a change in weather. It requires adjustment, not panic. What Publishers Are Really Being Asked to Rebuild The search story is often described as a traffic story, and that is true on the surface. Fewer clicks from search means fewer accidental arrivals, fewer advertising impressions, and more pressure on the economics of journalism and media. But there is another layer beneath it. Publishers are being asked to rebuild direct relationships with readers. For years, search partially handled discovery. A reader did not need to remember where to go. A search engine could route them there. That system rewarded visibility, optimization, and the ability to answer a query well enough to appear near the top. It also allowed publications to depend, at least partly, on audiences they had not truly earned as loyal readers. Lynch’s comments suggest that this dependency is becoming less stable. Search Engine Journal reports his view that strong authoritative brands and small loyal niche publications may be better positioned, while broader middle-positioned brands face more pressure. He called this a “barbell effect.” ( Search Engine Journal ) There is something revealing in that structure. The future may reward publications that feel less like interchangeable content suppliers and more like places with memory. Places readers return to directly because trust has accumulated over time. Places whose authority does not depend entirely on being surfaced by someone else’s interface. This is not a return to the old web. It is a reminder that trust has to live somewhere. When search sends fewer people outward, trust must be carried more directly by the publication, the brand, the community, and the reader’s own habits of return. When AI Enters the Conversation The Threads story brings the same shift into a more intimate space. Search is where we ask the world. Social platforms are where we ask each other. When Meta AI enters Threads, the assistant is no longer only a separate tool waiting in another app. It becomes part of the conversational layer. A user can bring it into a public exchange for an answer, a clarification, or additional context. In many ordinary situations, that could be useful. Online conversations often contain moments where people want a definition, a source, a summary, or a shared reference point. The issue, according to Engadget’s reporting, was not only the existence of the tool. It was the boundary around it. The @ meta.ai account was visible to everyone, even though most users could not interact with the bot yet, and users discovered that they could not block it like a regular Threads account. Meta told Engadget that users could manage exposure by muting, using “not interested,” or hiding replies to their own posts if someone brought the AI in. ( Engadget ) That may be enough from a product-control perspective. It is not quite the same from a human perspective. A conversational tool does not need to be frightening to require consent. A system can be useful and still need boundaries. In fact, the more useful it becomes, the more carefully those boundaries need to be designed. The Shift Beneath Both Stories The synthesis is not that AI is destroying search or that platforms are forcing people into a future they did not choose. That reading is too flat. The more interesting movement is subtler. The web is shifting from navigation to response. In publishing, that means fewer journeys from search results into original sources, and more pressure on publishers to build direct audiences. In social media, it means AI moving from a separate destination into the flow of conversation itself. In both cases, information arrives closer to the point of need. That closeness can be genuinely helpful. It can reduce effort. It can make information easier to access. It can support people who are tired, overloaded, disabled, time-constrained, or simply trying to understand something quickly. A more responsive web is not automatically a poorer web. But response changes the role of context. When an answer comes quickly, the system carries more responsibility for what has been compressed. When a summary appears before the source, the interface shapes what the user notices. When AI enters a conversation, the platform decides how visible, optional, and governable that presence feels. The question is not whether response is useful. It clearly can be. The question is whether response can preserve proportion. The Value of Useful Friction In design culture, friction is usually treated as something to remove. Every extra click is seen as a possible leak. Every delay becomes a failure. Every pause is optimized away. Often, that instinct improves things. Nobody needs confusion for its own sake. Nobody benefits from bad navigation, buried settings, or unnecessary complexity. But not all friction is waste. Some friction is attention. The click into a full article. The visible source beneath a claim. The second opinion. The extra minute spent realizing that an answer is more complicated than the first summary suggested. The ability to leave one platform, follow a path outward, and return with a wider view. These are not nostalgic rituals. They are part of how context enters the mind. A zero-click environment may be efficient, but it must find another way to make context visible. A social AI assistant may be useful, but it must find a clearer way to respect the boundaries of the room it enters. In both cases, the design problem is not speed. It is trust. Response as Responsibility A responsible response shows where it came from. It preserves the possibility of depth. It allows refusal. It makes boundaries legible. It does not confuse convenience with care. That is the principle connecting these two stories. Condé Nast preparing for a low-search future tells us that publishers may need to become destinations again, supported by direct relationships and loyal audiences rather than accidental discovery alone. Threads testing Meta AI tells us that platforms are beginning to place AI inside the social spaces where meaning is negotiated between people. Both developments point toward an internet that answers faster, closer, and with less movement required from the user. The opportunity is real. So is the need for care. When I think about “zero-click,” I do not hear only a publishing metric. I hear a cultural metaphor. A web with fewer clicks may also become a web with fewer visible paths. When the paths become less visible, trust has to be carried differently: by the interface, by the source, by the product choices, by the institutions behind them, and by the habits we keep as readers and participants. The work ahead is not to defend the old web exactly as it was. The old web had its own noise, manipulation, and fatigue. The work is to preserve the habits of attention that helped people become more discerning inside it. To ask where an answer came from. To notice when a tool enters a conversation. To build direct relationships with sources worth returning to. To keep enough space between question and answer for understanding to breathe. The internet is learning to answer back. The question now is whether we can teach it to answer with proportion, and whether we will remember that sometimes understanding still requires us to move. --- ## Media (/media) - Full Record ### Press **Revolutionizing Emotional Wellness: Innovator Manu Rehani Unveils Issa's Groundbreaking Emotion-Aware AI** Source: Texas Squared (startup newsletter) URL: https://marc1919.medium.com/texas-squared-startup-newsletter-449-ec8316420677 This newsletter mention presents Issa as an emotion-aware AI platform focused on emotional wellness and human support. It reinforces the broader positioning of Manu's work around AI systems that can interpret and respond to human emotional signal more intelligently. **How Companies Can Retain Employee Trust During the AI Revolution** Source: InformationWeek URL: https://www.informationweek.com/machine-learning-ai/how-companies-can-retain-employee-trust-during-the-ai-revolution This article frames AI adoption as a trust challenge inside organizations, especially as employees worry about job security and the pace of automation. It focuses on how enterprise leaders can preserve confidence, transparency, and morale while still taking advantage of AI's business potential. **3 AI Startups Tackle M&A's Biggest Blind Spots From Different Angles** Source: Forbes (Jennifer J. Fondrevay) | Date: 2026-06-30 URL: https://www.forbes.com/sites/jenniferjfondrevay/2026/06/30/3-ai-startups-tackle-mas-biggest-blind-spots-from-different-angles/ This Forbes article profiles three AI-powered platforms making leadership and cultural compatibility measurable before an M&A deal closes. It details how Humanaq's Execution Alignment Index applies Manu Rehani's Cognitive NLP to surface execution risk during due diligence, alongside Grodivo's standardized culture assessment and NayaDaya's post-merger emotion analytics. --- ### Interview **Emotion Aware AI** Source: Silicon Hill News (podcast) URL: https://ideastoinvoices.libsyn.com/manu-rehani-founder-of-issa This podcast profile describes Manu as an innovator in applied behavioral and cognitive linguistics, with work spanning emotion-aware AI, GPTs, LLMs, mental health, and wellness applications. It also ties his current work to his broader background as an inventor, mentor, board advisor, founder, and patent holder. **AI, Technology, and Spiritual Awakening** Source: YouTube (video discussion) URL: https://youtu.be/8YjaYvYYLwk?si=reTeX0MeUXQJYmRo This video explores the unexpected intersection of artificial intelligence, spirituality, relationships, and emotional maturity. It frames AI not only as a technological force, but also as a possible catalyst for deeper reflection on human belief, connection, and inner development. **Mental Health Champions: Why & How Manu Rehani of Hey Issa Is Helping To Champion Mental Wellness** Source: Authority Magazine (Eden Gold) URL: https://medium.com/authority-magazine/mental-health-champions-why-how-manu-rehani-of-hey-issa-is-helping-to-champion-mental-wellness-8c48668dcac7 This interview highlights Manu's work at the intersection of AI, emotional wellness, and mental health support. It presents Hey Issa as part of a broader effort to use emotionally aware technology to help people build healthier relationships with themselves and others. **How AI Is Disrupting Our Industry, and What We Can Do About It** Source: Interview with Cynthia Corsetti URL: https://www.cynthiacorsetti.com/manu-rehani-of-heyissa-how-ai-is-disrupting-our-industry-and-what-we-can-do-about-it/ This interview explores how AI is disrupting industries and what leaders, founders, and professionals can do to adapt. It positions Manu's work with HeyIssa within the broader shift toward applied AI, behavioral intelligence, and human-centered technology. --- ### Speaking **Inventing at the Edge: Systems Thinking from 12 Patents** Maps to: Systems Complexity Drawn from 12 patents spanning semiconductor manufacturing, infrastructure, and AI-native architectures - a practitioner's view of how behavioral and systems thinking meet hard technical problems that simulation alone cannot resolve. **Decision Architecture Under Pressure** The behavioral mechanics of high-stakes decisions: how organizational structure shapes judgment, where human and machine intelligence complement each other, and what breaks when the system is under stress. **The AI-Native Renaissance** The creative, seeking spirit that AI is about to unlock at global scale - and what leaders, builders, and investors need to do now to make sure that potential is not wasted. **Inventing at the Edge: Lessons from 12 Patents** A practitioner's account of innovation at the frontier of cloud, AI, wearables, and autonomous systems - and what the patent process teaches about disciplined creativity. **The Relevance Fit Framework** Why product-market fit is a floor, not a ceiling. The four-dimension relevance test: emotional, social, spiritual, and environmental fit - and how teams that use it build things that last. **Behavioral Intelligence in the Age of AI** How behavioral science reshapes decision-making when AI enters the room - and why the human signal matters more, not less, as automation increases. --- ### Advisory **Frontier Systems Research** Best for: Applied R&D, experimental systems, scientific modeling, and technical exploration Exploring how complex systems behave under real-world constraints - from semiconductor manufacturing and infrastructure durability to AI-native architectures, simulation systems, and emerging scientific tools. **Investor Intelligence** Best for: GPs and allocators assessing AI-native opportunities Reading the behavioral and organizational signal underneath an AI-native pitch: whether the team has the decision architecture to compound, and whether the founder can hold gumption under pressure. **Relevance Fit Strategy** Best for: teams designing products, ventures, or organizational change Stress-testing whether the work actually resonates - emotionally, socially, psychologically, and environmentally - with the people it touches. Product-market fit is necessary. Relevance fit is the harder, more durable bar. **Leadership Decision Architecture** Best for: CEOs, boards, and executive teams navigating high-consequence change Who decides what, with which evidence, under what constraints - and what happens when the machine is confidently wrong. **Behavioral Intelligence Systems** Best for: founders embedding AI into decision and product systems Designing the interface between human judgment and machine output so each stays in its proper lane. The behavioral work is reversing the common error of trusting AI where it is confident and doubting humans where they are uncertain. --- ### patent **Autonomous Systems Decision Framework** Source: USPTO | Date: 2019 Decision logic and uncertainty handling in autonomous systems navigating dynamic real-world environments. **Wearable Biometric and Behavioral Monitoring System** Source: USPTO | Date: 2018 Continuous behavioral and biometric monitoring via wearable devices, bridging physiological signals and behavioral inference. **Contextualization Equations for Language Models** Source: USPTO | Date: 2016 URL: https://patents.google.com/patent/US9269353B1 The mathematical framework for contextualizing language inputs - sitting beneath today's large language models. **Data Awareness Protocol for Distributed Systems** Source: USPTO | Date: 2017 URL: https://patents.google.com/patent/US9787454B2 Cloud storage and data streaming awareness protocols - one of the foundational inventions behind cloud infrastructure as we know it. --- ## Documents **The Bifurcation of the Word** 2016 Leiden dissertation on orthographic schisms, the "Histree" of NLP, and the impending computational bottleneck - anticipating the shift from brute-force sequence models toward embodied, cognitive architectures. Category: dissertation **Forbes - 3 AI Startups Tackle M&A's Biggest Blind Spots From Different Angles** Forbes profiles three AI platforms making leadership and cultural compatibility measurable before an M&A deal closes - including how Humanaq's Execution Alignment Index applies Manu Rehani's Cognitive NLP to surface execution risk during due diligence. Category: press --- ## Sitemap https://rehani.co/sitemap.xml