Overeducated, Ph.D.

Writings, Regrets, and Re-skillings during the AI Revolution

Author: Professor Lamb

  • The Knot in the Network

    The Knot in the Network

    In Computer Networking, the basic unit of science is the ‘node.’ All computers and chips, anything that sends or receives traffic, is a node. Nodes are connected by links. Networks consist of nodes and links. Some nodes are routers, some are hosts.

    The origin of the word is the Latin nodus, “knot”—the thing you tie in a rope. Trace it further back to its Indo-European root ned-, “to bind or tie,” which also produced Latin nectere, “to bind,” which gives us connect, annex, and nexus; and through the Germanic line it produced the English word net. “Network node” is therefore very nearly a redundancy—the knot of the knotted thing!

    For most of its career the word named distinguished points in continuous things, such as where orbits cross in astronomy. But for Computer Networking, the decisive precedent is electrical. In 1845 Gustav Kirchhoff, still a student at Königsberg, published the two circuit laws every engineering student now learns; the current law—what flows into a junction flows out—is known in German as the Knotenregel, the knot rule, because a junction where conductor branches meet is a Knoten, a knot. Electrical engineering was calling its junctions knots a century before anyone switched a packet.

    Königsberg, fittingly, is also where network mathematics comes from: Leonhard Euler’s 1736 analysis of the city’s seven bridges is the conventional origin of graph theory, the abstraction (a graph = points joined by lines) by which orbits, stems, circuits, and telephone exchanges become the same object. The word graph itself came later, coined by the mathematician James Joseph Sylvester in 1878; mathematicians tend to say vertex, engineers say node, and they mean the same thing.

    Computer networking inherited both lineages at once. Paul Baran’s 1964 RAND studies, On Distributed Communications, asked whether a web of switching nodes with redundant links could survive the loss of many of them. (Spoiler: YES).

    When the ARPANET went live in 1969, its design split the graph world into two: a communication subnet built of Interface Message Processors were called the “nodes,” with UCLA’s IMP as node number one—versus the timesharing computers that actually served users, the “hosts.” That division of labor is why the network’s very first Request for Comments, RFC 1 of April 1969, is titled simply “Host Software.”

    The standards later fixed the usage into law. RFC 8200, the current IPv6 specification, defines a node as a device that implements the protocol, a router as a node that forwards packets not addressed to itself, and a host as any node that is not a router. So in IETF terms every host is a node, and routers are the nodes that aren’t hosts.

    The IEEE, meanwhile, preserves the older topological sense: the 802 LAN standards largely avoid node for end systems, preferring station, while IEEE glossary usage keeps the graph-theoretic meaning — an endpoint of a link, or a junction common to several. The definitions differ because they draw different graphs: to Ethernet, a switch is a node; to IP, the whole switched segment collapses into a single link and the switch vanishes.

    Two thousand years of usage agree on the essential point. To call something a node is to make a relational claim: a node is not a kind of thing but a kind of position. A device becomes a node the moment it is tied into something larger, and ceases to be one the moment it is cut loose. The Romans would have found nothing strange here — a nodus was always a place where separate strands are made to hold. We did not stretch the word to fit our machines; we finally built the thing the word was waiting for.

  • Four Rules for Governing World Models

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 7 of 7.

    Critical Engineering Recommendations

    Require independent safety and quality testing before deployment. A world-model robot controller should not be used around workers until an independent evaluator has tested the exact model, robot, task, and factory environment. Testing should include unusual parts, blocked sensors, lighting changes, worker entry into the cell, damaged equipment, and other foreseeable failures. A successful vendor demonstration geared toward marketing and the venture capitalists is not enough. The system should also be retested after major model updates, new training data, or changes to the environment. If the system cannot show safe behavior under these conditions, it should not be used in production environments.

    Protect workers, jobs, and the right to stop the system. Employers should complete a risk and impact assessment before introducing this type of automation. The assessment should estimate jobs lost, jobs changed, wages affected, training needs, and new safety hazards. Workers and their representatives should participate before deployment, not after the purchasing decision has already been made. Factories should have a clearly tested emergency-stop process, and workers should have the authority to halt operation without retaliation. Employers should also fund retraining, continued manual practice, and transition support for workers whose jobs are eliminated or reduced. Productivity gains should not be treated as a complete benefit calculation while job loss and lost wages are externalized by the company.

    Do not count a human monitor as a safety control. The most common industry answer to the risks described above is to place a trained person in a supervisory position and treat that person as the safeguard. The Avride investigation is direct evidence that this does not work. A qualified safety operator occupied the driver’s seat during all sixteen crashes under review, and in only one of them did that operator attempt to intervene. A learned model/agent acts faster than a passive observer can reliably react, and a supervisor watching a system that behaves correctly most of the time will stop watching it closely. Employers and regulators should therefore require that a deployment be safe on its engineered limits alone, e.g., force and speed ceilings, protective stops, physical separation, and restricted task boundaries that hold whether or not anyone intervenes. Where human intervention is still claimed as a mitigation, it should be measured rather than assumed: the assessment must document the time a worker actually needs to detect a fault and stop the machine, verified under realistic conditions rather than during a scheduled test. Supervision should be treated as a way to catch problems and improve the system over time, never as the primary barrier between a machine-learning model and a human body.

    Assign responsibility and blame before an accident. Contracts and regulations should identify which party is responsible for the foundation model, robot integration, safety controls, software updates, and day-to-day operation. Each deployed version should have a model data provenance card that records its origin, training changes, known limits, and safety tests. Serious failures and near misses should be reported to a shared public database so that one company’s accident can prevent another company’s possible fatality. Regulators should also require secure update procedures and checks for compromised model weights because research has shown that generative pipelines can be poisoned through hidden backdoors.

    World models may eventually improve reliability, but that possibility does not justify testing their emerging capabilities on workers in the field as the WEF suggests. Our policy goal should not be to stop world-model research. It should be to prevent companies from shifting the risks of experimental automation onto people who did not design the system and may not share in its profits. Independent benchmarks and quality testing, enforceable worker protections, job-impact planning, public incident reporting, and clear liability laws would require the evidence to catch up with the runaway marketing and hype. Until tech companies and their capitalist advocates can demonstrate that these systems are safe in the real-world physical environments where they will operate, they should not be permitted to gamble with workers’ jobs, property, or lives.

  • The Policy Problem of Physical AI

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 6 of 7.

    Introduction and Context

    World models are AI systems that try to predict how the physical world will change. In a factory, that prediction can become a robot action. This makes the technology more dangerous than an ordinary generative AI system. A bad chatbot answer may ‘hallucinate’ or confuse someone. A bad robot command may crush a hand, damage equipment, or kill someone. FLUX-mimic is already being tested at Audi, even though its reliability has not been independently established. The main policy problem is therefore simple: companies are moving experimental AI from video generation into industrial control before workers, regulators, and the public have reliable ways to judge whether it is safe.

    Relevance to Policy and Practice

    Policy action is needed because the costs and benefits will fall on different groups. Audi and its vendors may gain faster production, lower labor costs, and a new market for foundation models. Workers may face injury, job loss, lower wages, or reassignment to monitoring roles with less skill and bargaining power. The safety risk is already real and urgent. NIOSH identified 41 robot-related deaths in the United States between 1992 and 2017, and it warns that unexpected contact, crushing, trapping, and concern about job loss remain important risks as robots work closer to people (CDC 2024). In May 2026, federal regulators opened a defect investigation into Avride, an Uber robotaxi partner operating in Austin and Dallas, after sixteen crashes in which the automated driving system merged into occupied lanes, failed to slow for traffic ahead, and struck objects partly blocking the road (O’Kane 2026). Interestingly, a trained human monitor sat in the driver’s seat for every one of them, and in only a single incident did that monitor attempt to take over. The regulator’s warning about “inappropriate assertiveness and insufficient competence” provokes the very question this report is concerned with: can a world model remain safe at the moment an ordinary reality stops resembling its training data?

    Current world models add another problem: their actions may be difficult to predict and control because the models learn stochastic patterns instead of following a complete set of explicit rules. Symbolic guardrails and explicit prohibitions at some point become necessary again to avoid worst-case statistical blunders. Public stakeholders should have the right to participate in the deliberation and articulation of such absolute prohibitions. These protections remain necessary, but a learned controller raises questions that ordinary guarding and emergency stops cannot answer. Regulators also need to know how the model was trained, how often it fails, what changes after fine-tuning, and whether it behaves safely outside its demonstration data. The NIST AI Risk Management Framework can help organizations identify and manage these risks, but it is voluntary and does not itself prevent an unsafe system from entering a factory (NIST 2023). In short, the current system relies too heavily on vendors and employers to judge their own technology.

    References

    CDC (Centers for Disease Control and Prevention). 2024. “Robotics in the Workplace: An Overview.” National Institute for Occupational Safety and Health, February 9, 2024.

    NIST (National Institute of Standards and Technology). 2023. Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. Gaithersburg, MD: National Institute of Standards and Technology.

    O’Kane, Sean. 2026. “Uber Partner Avride Is under Investigation for Self-Driving Crashes.” TechCrunch, May 8, 2026.

  • When a World Model Makes the Wrong Physical Decision

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 5 of 7.

    Risks of World Models

    Risks of deploying world models at scale in physical robotics include possible death, injury, and damage. A robot controlled by a world model can make the wrong physical decision, injure or kill a worker, damage equipment, or stop a production line. The same system may also eliminate jobs or reduce workers to monitoring a process they no longer fully understand. These risks are harder to manage because FLUX-mimic is still a vendor-led production test, not an independently validated industrial control system. The critical question is therefore not whether the model can produce convincing demonstrations (it can). It is whether the system is safe, reliable, and accountable when something unexpected happens around real people and machinery.

    The most immediate engineering risk is unreliable behavior. A world model can produce an answer that looks reasonable while being physically wrong (WEF 2026). The literature survey in the Technical Summary section already showed that current video models can generate realistic scenes that fail the fundamental laws of gravity, fluids, and heat (Ding et al. 2025). That problem becomes far more serious when the output is a robot command instead of a video frame. A bad prediction could cause a robot to grip too hard, move into a worker’s space, drop a component, or damage a vehicle. Industrial robots have caused fatal accidents before, especially during maintenance, testing, setup, and other situations when workers enter the robot’s operating area (CDC 2024).

    Another possible risk is job loss. The main economic purpose of this system is to automate assembly work that previously required people. If it succeeds, some repetitive and injury-prone jobs may disappear, but so may stable manufacturing jobs that support workers and communities. The benefits will not be shared automatically. Audi and its technology vendors may gain productivity and lower labor costs while workers face layoffs, reassignment, or pressure to accept lower-skill monitoring roles. Retraining may help some workers, but it is not a complete answer if there are fewer comparable jobs available. Any serious evaluation should therefore count displaced workers and lost wages as costs, not treat them as side effects outside the engineering problem.

    Automation can also weaken human oversight over time. Experienced workers often notice small changes in sound, motion, resistance, or part quality before a formal alarm appears. If people stop performing the task, they may gradually lose the practical knowledge needed to recognize when the robot is behaving dangerously. Supervisors could then become passive monitors of a system they cannot fully inspect or explain. For human oversight to mean anything, workers must retain the authority to stop the robot, receive training on its known failure modes, and regularly practice the manual skills needed to take over safely.

    Finally, the number of supply-chain partners can create a diffusion of irresponsibility effect when a serious failure occurs. Black Forest Labs trained the foundation model FLUX 3, mimic adapts it for robot control via FLUX-mimic, Audi places it on the factory floor, and workers are exposed to the result. If the robot injures someone, each organization may point to another part of the supply chain. Open-weight releases may make this problem worse because many companies can create modified versions with different data, safeguards, and quality controls. Research has also shown that generative pipelines can be deliberately compromised through backdoors and data poisoning (Lapid and Dubin 2025). Clear responsibility, documented model changes, independent safety testing, and incident reporting should therefore be minimum conditions for deployment—not add-ons added after a fatal accident occurs.

    References

    CDC (Centers for Disease Control and Prevention). 2024. “Robotics in the Workplace: An Overview.” National Institute for Occupational Safety and Health, February 9, 2024.

    Ding, Jingtao, Yunke Zhang, Yu Shang, Yuheng Zhang, Zefang Zong, Jie Feng, Yuan Yuan, et al. 2025. “Understanding World or Predicting Future? A Comprehensive Survey of World Models.” ACM Computing Surveys 58 (3): Article 57.

    Lapid, Raz, and Almog Dubin. 2025. “Backdoors in Conditional Diffusion: Threats to Responsible Synthetic Data Pipelines.” AAAI 2026 Workshop on Shaping Responsible Synthetic Data in the Era of Foundation Models.

    WEF (World Economic Forum). 2026. Top 10 Emerging Technologies of 2026. Geneva: World Economic Forum, pp. 29–31 and 47–48.

  • A World Model Enters the Factory

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 4 of 7.

    Use Case Description

    For a recent use case analysis, timely as of summer 2026, I chose FLUX-mimic, a new world-model robotic system undergoing production testing at Audi factories for the installation of flexible door seals and the handling of car components in final vehicle assembly (Black Forest Labs 2026b). The system is useful for this report because its learned video-action policy is derived from a multimodal foundation model trained on large-scale video, vision, language, and robotics data rather than being specified entirely through task-specific rules (Black Forest Labs 2026a). That learned policy is only one component of a complete robot-control system that also includes sensors, middleware, action decoding, actuators, conventional software, and safety controls. The relevant engineering question is therefore not whether the model acts alone, but whether the entire robot cell remains safe when the learned component makes an incorrect prediction.

    Black Forest Labs’ wager in its public announcement “Real World Models” is that a generative-AI world model trained to represent reality convincingly must implicitly learn some of reality’s physical regularities, such as how a cable droops or falls, and that this video-learned knowledge can be redirected through transfer learning from pixel generation to motor control. That places the system squarely in the category that the World Economic Forum’s 2026 report describes as world models, alongside the NVIDIA Cosmos platform it highlights (WEF 2026). Adopting Ding et al.’s taxonomy, FLUX-mimic is a future-prediction world model derived partly from generative video AI.

    The main use actors are as follows: Audi’s manufacturing engineers select candidate workflows that have not previously been automated, such as tasks involving floppy or deformable parts, in which geometry is not fixed and strictly repeated motion usually fails. Robotics engineers then fine-tune the pretrained FLUX 3 foundation model into a “video-action model” for each selected task using human demonstrations. The vendors describe this adaptation as requiring only a limited amount of task-specific demonstration data rather than extensive conventional programming (Black Forest Labs 2026b). The adapted model is deployed to a robot cell, where it consumes camera streams and outputs control actions at operational speed. Audi’s process engineers then deploy the cell on the production assembly line, while line technicians supervise operation and handle exceptions. Behind this sequence stands an under-discussed fourth actor: the pretraining pipeline and the broader media economy that supplied its data. Black Forest Labs describes training at massive scale but does not fully disclose the provenance, licensing status, or consent conditions governing that material. This opacity creates unresolved copyright and labor concerns, particularly because the same company sells video generation into film and advertising markets that may disrupt work traditionally performed by human crews and performers.

    Benefits

    The clearest near-term benefit is the automation of physical tasks that conventional industrial robots handle poorly. Traditional systems excel when parts are rigid, locations are fixed, and motions can be repeated exactly; they are far less capable with seals, cables, fabric, food, and other deformable or fragile materials whose shape changes during handling. A world-model controller that predicts those changes could extend automation into final-assembly work that remains manual, repetitive, and ergonomically difficult. In Audi’s use case, successful seal installation could therefore produce both a productivity benefit and a worker-safety benefit by reducing sustained force, awkward posture, and repetitive motion provided that the new robotic system does not introduce greater hazards of its own. A related benefit is faster adaptation and safer experimentation. Conventional automation may require new fixtures, task-specific programming, and extensive validation whenever a product or workflow changes. FLUX-mimic’s vendors claim that its controller can instead be adapted quite quickly compared to traditional robotics (re)programming.

    The possible impact also extends well beyond manufacturing. In transportation, world models could help vehicles anticipate how traffic participants and road conditions may evolve, compare possible maneuvers, and rehearse uncommon emergencies before deployment. In medicine, related predictive models could support treatment planning, medical imaging, surgical rehearsal, or assistive robotics by estimating how a patient or procedure may respond under different interventions. In hazardous-materials handling and disaster response, robots could explore contaminated, explosive, radioactive, or structurally unstable environments while keeping people farther from immediate danger. World models have also been proposed for climate and weather scenarios, scientific modeling, and other systems in which direct experimentation is costly or impossible (Ding et al. 2025). Across these fields, the common benefit is not autonomous “understanding” in the strong sense, but the ability to compare possible futures before choosing an action in the real world.

    References

    Black Forest Labs. 2026a. “FLUX 3 — Real World Models.” Black Forest Labs, July 23, 2026.

    Black Forest Labs. 2026b. “FLUX 3 x mimic: The Next Generation of Video-Action Models.” Black Forest Labs, July 23, 2026.

  • Plausible Physics Is Not Physical Understanding

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 3 of 7.

    Assessment and Analysis

    Read critically, Ding et al’s most consequential contribution is the evidence it compiles against the field’s own marketing and hype. The product category is named for understanding the world, yet the authors document generative AI systems failing precisely at physical reasoning while succeeding at visual plausibility. That gap is diagnostic of what these systems optimize: a video generator is trained to make the next frame look right to a human eye, so gravity, friction, and thermal behavior are learned only insofar as they leave visible traces in pixels. Internal coherence is rewarded; correspondence to the world is, at best, a by-product. This gap could produce a new category of physical AI failure: not merely videos with broken physics, but robots whose plausible-looking predictions generate unsafe behavior. A hallucination on a factory floor, not to mention in a school, hospital, or road environment, may cost lives.

    Their own evidence also licenses a critique of the survey’s own central move. Ding and colleagues resolve the field’s definitional dispute by declaring understanding and prediction to be two functions of a single underlying idea. It is an elegant synthesis, but I think it is too slippery still. Current laboratory studies and frontier demonstrations suggest that the two functions come apart in practice: a system can be excellent at generating plausible futures while holding no rule-governed model of the present. If that is so, then ‘understanding’ and ‘predicting’ are not two aspects of one capability but two distinct claims. The shared slippery label permits a vendor to demonstrate the second while marketing the first.

    The survey’s “missing benchmarks” observation compounds this problem, and for the developer audience it is arguably the paper’s most important point: deployment of world-model systems is proceeding without an agreed method for validating opaque learned controllers across distribution shifts, model updates, and changing operating conditions. Industrial robotics is not devoid of safety governance; existing machinery standards address robot cells, guarding, protective functions, integration, and foreseeable human access. The unresolved gap is narrower but consequential: these instruments do not yet provide a canonical, model-specific method for establishing whether a learned controller remains reliable after fine-tuning, software updates, sensor drift, or exposure to conditions outside its training data. Without such evidence, a vendor demonstration can become a de facto performance test whose task, conditions, and footage are selected by the vendor.

    Two further limitations of the survey should be named. As a survey, the paper produces no original experiment; its findings are curated from others’ work, and its authority rests on the quality of its taxonomic synthesis rather than on evidence it generated itself. In a field moving this quickly, any survey’s shelf life is also short, and this one already predates several systems now central to public discussion, including the FLUX 3 lineage I take up in the next installment, my Use Case Analysis.

    Ding and colleagues provide definitional clarity and the taxonomy that discussion of world models has so far lacked. The state of the field they document nevertheless remains premature: researchers have demonstrated potential, but not safety or reliability, and quality-oriented benchmarks remain fragmented. Once again, Silicon Valley’s “move fast and break things” ethos is entering a high-impact domain with physical consequences before civil society and government institutions have developed adequate model-specific assurance and accountability mechanisms.

    Reference

    Ding, Jingtao, Yunke Zhang, Yu Shang, Yuheng Zhang, Zefang Zong, Jie Feng, Yuan Yuan, et al. 2025. “Understanding World or Predicting Future? A Comprehensive Survey of World Models.” ACM Computing Surveys 58 (3): Article 57.

  • What Is a World Model?

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 2 of 7.

    Technical Summary

    A. Introduction

    In “Understanding World or Predicting Future? A Comprehensive Survey of World Models,” Jingtao Ding and eleven colleagues at Tsinghua University (published in ACM Computing Surveys 58, 3, 2025) provide the first comprehensive peer-reviewed survey of world-model research. Its central move is to resolve the term’s contested definition by organizing the entire literature around two functions: i. building internal representations that help a machine understand the world and ii. predicting future states of the world to guide decisions.

    B. Description

    The problem that Ding et al. address is that “world model” has come to mean different things to different research communities, all hiding beneath the marketing hype. One lineage, descending from Ha and Schmidhuber’s 2018 work and from model-based reinforcement learning, treats a world model as a compressed internal representation that an agent consults to make decisions. Ha and Schmidhuber used learned latent representations to produce the first reported solution of the CarRacing benchmark. In a separate VizDoom experiment, they trained a policy entirely inside an environment generated by the model and transferred that policy successfully to the original game (Ha and Schmidhuber 2018).

    A second lineage, energized by video generators such as Sora and FLUX, treats a world model as a simulator that renders plausible future states of the physical world. The “Joint Embedding Predictive Architecture,” a.k.a. JEPA, proposed by Yann LeCun, former Chief AI Research Scientist at Meta, argues that a machine should both represent the world abstractly and use that representation to imagine outcomes before acting. The embedding in that name points to a mechanism both lineages increasingly share: instead of learning only from text data or reproducing every image pixel, many current systems compress video, lidar, motion, and other sensor data into a ‘latent’ or abstract space, which consists of numerical vector representations intended to preserve important event information and dynamic relationships while omitting overwhelming raw detail. Prediction then happens in that compressed space rather than at the pixel level, which is what makes the approach computationally tractable at scale, but also what makes its omissions consequential.

    The authors conduct a systematic review in three passes. First, they survey the internal-representation branch: world models inside model-based reinforcement learning agents, and the more recent finding that large language models (LLMs) appear to acquire latent world knowledge, including spatial and temporal structure, without being explicitly trained for it. Second, they survey the future-prediction branch, tracing a progression from video generation toward interactive embodied environments, from producing footage of the world to producing worlds an agent can act in. Third, they examine three application domains (autonomous driving, robotics, and social simulacra), showing that each domain draws on their two functions in different proportions. The authors attempt to unify these two fields as two functions of one underlying idea: understanding the present and predicting the future.

    The paper closes with an open problem and findings that are largely negative. Current video generators achieve visual realism but fail systematic tests of physical law: the authors cite diagnostic studies showing errors in gravity, fluid, and thermal dynamics, and evidence that scaling produces case-based rather than rule-based generalization. So far world models ‘memorize’ situations without inducing the physics beneath them. Finally, the authors document that evaluation is fragmented: because the field pursues divergent goals with heterogeneous methods, no canonical benchmark or metric for “being a good world model” yet exists.

    References

    Ding, Jingtao, Yunke Zhang, Yu Shang, Yuheng Zhang, Zefang Zong, Jie Feng, Yuan Yuan, et al. 2025. “Understanding World or Predicting Future? A Comprehensive Survey of World Models.” ACM Computing Surveys 58 (3): Article 57.

    Ha, David, and Jürgen Schmidhuber. 2018. “Recurrent World Models Facilitate Policy Evolution.” In Advances in Neural Information Processing Systems 31 (NeurIPS 2018), 2450–2462.

  • World Models: The Next Frontier of AI Hype

    Series note. To mark the completion of my ECE 5864 Critical Engineering capstone, I’m publishing a seven-part adaptation for a special week of ITE 135: AI Awareness. These posts are longer and more formal than my usual entries because they began as a graduate research paper. This is Part 1 of 7.

    Introduction

    World models are AI models that form—through massive and deep machine learning on video, vision, language, and robotics data—an internal representation of a physical environment’s state and dynamics to predict how that environment may change and evaluate possible actions within it. They are a key enabling approach for an emerging generation of physical-AI systems and robotics. They are also the latest frontier of Silicon-Valley-inspired hype and market speculation because they could enable so-called ‘agentic AI’ to control autonomous machines in real-world environments by making inferences, selecting actions, and predicting their effects. If the technology proves viable at scale, world models could enable robots, cobots, autonomous vehicles, and other physical-AI systems to replace some human labor or better collaborate with human workers. The potential impact is immense but remains largely speculative at present.

    Last month, the World Economic Forum selected world models as one of its Top 10 Emerging Technologies of 2026. Its report presents them as a foundation for AI systems that can plan, reason about, and act autonomously in physical settings, including robotics, industrial operations, climate modeling, and transportation (WEF 2026, 29–31). Unfortunately, their business-facing report glosses over a central critical-engineering risk: a world model may be coherent within its own representation yet still grossly misunderstand the world, wreaking havoc and harm in the environment. Critical errors and ‘bug’ defects may remain hidden until a world model is actually deployed and encounters real people/places/things outside its training environment.

    World models are an active area of research, development, and technology investment. Major engineering obstacles remain, including model degradation over time, stringent reliability requirements for systems operating around people and property, computational efficiency at scale, and high energy and environmental costs. Terminological confusion and a proliferation of hype-driven buzzwords further complicate analysis and assessment. None of these uncertainties has dampened enthusiasm: Silicon Valley is wagering heavily and publicly that world models will become both deployable and lucrative in the near future.

    This critical-engineering report is written for developers, technology managers, workers, and other public stakeholders who may be affected by or curious about emerging world-model and physical-AI systems. In the next installment, I summarize the most comprehensive peer-reviewed survey to date of the field’s competing definitions, main technical approaches, major application domains, and unresolved problems. I argue that “world model” currently functions as both a technical category and a marketing device. Vendors can slide rhetorically from a model’s ability to generate plausible representations to claims that it understands physical reality well enough to act safely in settings such as schools, malls, and elder care. Although world models show substantial promise for representing, predicting, and controlling spatiotemporal environments, limited demonstrations do not establish the calibrated reliability and accountability required in such settings. Given what is at stake in the natural and social world, these systems demand critical scrutiny before broad deployment.

    References

    WEF (World Economic Forum). 2026. Top 10 Emerging Technologies of 2026. Geneva: World Economic Forum, pp. 29–31 and 47–48.

  • Why Genealogy?

    Genealogy is the history of things that claim to have no history. It takes something that presents itself as natural, timeless, or unified—a value, an institution, a category of person—and shows it to be an assemblage built piecemeal out of contingent struggles, whose current meaning is a late capture rather than an original purpose. Standard histories narrate the development of given subjects. Genealogy asks how those subjects came to be given in the first place. And it writes in order to unsettle the present, not to commemorate the past.

    Friedrich Nietzsche (1844–1900), the German philosopher who started out as a professor of classical philology, coins the modern sense in On the Genealogy of Morality (1887). His target is morality itself—guilt, conscience, altruism, the ascetic ideal—things his contemporaries treated either as eternal or as tidy products of usefulness. His question is not “what is good?” but “how did good get made, by whom, against whom, and what is this value worth for the life that holds it?” qua an inquiry into the value of values. Nietzsche’s genealogy is openly a polemic. He writes it to diagnose the present — European nihilism, the sick animal called man — and to prepare a revaluation. It is history in the service of a fight, and it says so.

    Michel Foucault (1926–1984), the French philosopher-historian who held a chair at the Collège de France he named “History of Systems of Thought,” turned this into a research program. His methodological manifesto is the 1971 essay “Nietzsche, Genealogy, History”; the worked examples are Discipline and Punish (1975, on the birth of the modern prison) and The History of Sexuality, volume one (1976). Foucault fastens on Nietzsche’s German vocabulary. Genealogy rejects the search for Ursprung, a pure origin where a thing’s essence was already present waiting to develop, in favor of Herkunft, i.e. descent (the tangled, accidental lineages, tracked through bodies and through what he calls gray, meticulous, patiently documentary archives), and Entstehung, i.e. emergence (a thing arises as an episode in a struggle, and whoever currently holds it holds it the way one holds captured ground). To this he adds the power/knowledge axis: the objects we think we simply know—madness, delinquency, sexuality, even the inner soul—are constituted by mundane techniques like timetables, examinations, case files, and confession, rather than discovered lying around. And he names the genre’s temporal direction:

    Genealogy is a history of the present.

    It starts from something contemporary whose self-evidence bothers you and traces the contingent assembly that produced it until it stops looking necessary.

    Now contrasting genealogy with ordinary history-writing comes down to four things:

    1. What counts as a unit. Biography, intellectual biography, and political history receive their objects ready-made—a person, an oeuvre, a nation, a war—and narrate their development; the object anchors the story. Genealogy makes the object’s very existence the puzzle: not the life of a delinquent, but how “the delinquent” became a kind of person one could be; not the biography of an author, but how “the author” became a function we attach to texts (Foucault has a famous essay on exactly that). Biography in particular presupposes the thing genealogy interrogates—a continuous self whose intentions explain outcomes. For Foucault, the modern individual with an examinable interior is itself a product of disciplinary practice: “the soul is the prison of the body.”
    2. The shape of time. Picture conventional history as a single confident arrow: origin, development, arrival—the present as destination, faintly justified by the journey. Genealogy is the opposite shape: a tangle of thin lines converging on the present from scattered points labeled accident, reversal, appropriation, forgotten struggle. History’s continuity and teleology versus Genealogy’s discontinuity and contingency: the present as one outcome among possible others, held in place by force and habit rather than destiny.
    3. The Posture. Professional historiography since Leopold von Ranke—the nineteenth-century German historian who gave the discipline its ideal of archival neutrality, showing the past “as it actually was”—aspires to a view from nowhere. Genealogy denies there is one. It is openly perspectival, written from a present stake, and its success is measured not by completeness but by whether it loosens the seemingly self-evident. Foucault liked saying the point was to show things weren’t as necessary as all that. This also marks the difference in function: political history often explains or legitimates (the nation’s story, the institution’s founding), while genealogy problematizes.
    4. The Evidence. Not the correspondence of great men nor their ideas, but regulations, drill manuals, architectural plans, confession handbooks—power’s small print.

    Two caveats to keep in mind. Exposing something’s “low origins” doesn’t by itself refute a value—that would be the genetic fallacy. Nietzsche’s real question is what a value does to the life that holds it, and Foucault’s is what contingency reopens. Relatedly, genealogy needn’t debunk at all: Bernard Williams (1929–2003), the British moral philosopher, wrote a vindicatory genealogy of truthfulness. The same method of genealogy is now deployed to show why a value deserves keeping.

  • Digital Literacy From Above and Below

    A student can fail a digital literacy assignment while being digitally fluent in a world the instructor cannot even see. The student may struggle to format a Word document, attach the correct file to an email, or organize a spreadsheet. At the same time, that student may move effortlessly through online communities governed by their own symbols, references, genres, reputations, and unwritten rules.

    The institution recognizes the first set of practices as digital literacy. It creates course objectives for them, purchases textbooks about them, measures them with rubrics, and eventually certifies that the student possesses them.

    The second set may remain nearly invisible. No institution designed it. No professor controls it. No credential recognizes it. Yet it is still learned, practiced, policed, and used to determine who belongs.

    One literacy comes from above. The others grow from below.

    The Autonomous Model

    In Literacy in Theory and Practice (1984), the British anthropologist Brian Street distinguished between autonomous and ideological models of literacy.

    The autonomous model treats literacy as a neutral, self-contained skill. A person either possesses it or does not. Once acquired, it can supposedly be transported from one situation to another and will produce generally beneficial effects: better reasoning, greater employability, economic development, and fuller participation in society.

    This is the model institutions find easiest to use. If literacy is a stable skill, it can be defined. If it can be defined, it can be divided into competencies. Competencies can become learning outcomes. Learning outcomes can be assessed. Assessments can produce scores, certificates, course credits, and categories of people.

    The autonomous model turns literacy into something an institution can administer. Digital literacy is frequently presented this way. Students are expected to acquire a standard collection of abilities that will supposedly travel with them into college, employment, and civic life. The applications may change, but the institution still imagines a transferable substance called “digital literacy” residing inside the individual.

    This model is useful. I make part of my living teaching within it. It is also incomplete.

    The Ideological Model

    Street’s ideological model begins somewhere else. Literacy is not merely a technical ability stored inside an individual. It is a social practice embedded in particular relationships, institutions, purposes, histories, identities, and distributions of power.

    That means literacy is plural and lowercase: many literacies, not The Literacy (i.e., autonomous model). The literacy used to complete a tax form is not the same practice as the literacy used to interpret a religious text, participate in a group chat, maintain a medical chart, write computer code, create a meme, or decode the emotional politics of an Instagram comment section. These practices may share tools. They may all involve reading and writing. But they do not operate according to one universal set of rules.

    The Netflix series Adolescence gives us a disturbing example of vernacular digital literacy. Detective Inspector Bascombe and the other institutional adults can see the emojis exchanged beneath social-media posts, but they cannot interpret what the symbols mean within the young people’s social world. Bascombe’s son has to explain the coded layer. The adults can recognize the characters on the screen. They are not technically unable to read them. What they lack is membership in the community that gives the symbols their meaning.

    An emoji does not carry one permanent definition. Meaning depends on who is using it, where it appears, what has already happened, and which community is reading it. A heart may be affection in one context, mockery in another, and membership in a coded ideology somewhere else. The scene dramatizes Street’s argument powerfully. Literacy is not simply decoding a visible sign. It is knowing what that sign does inside a particular social relationship.

    The adults’ illegibility is also partly functional. Vernacular codes help communities recognize insiders and exclude outsiders. Young people continually create language that parents, teachers, police officers, employers, and algorithms cannot immediately interpret. If the adults master it, the community modifies it, abandons it, or moves elsewhere. This does not make vernacular literacy innocent or inherently progressive. A community-generated literacy can sustain friendship, creativity, humor, and mutual aid. It can also transmit misogyny, racism, conspiracy theories, cruelty, and violence.

    The Literacies an Institution Can See

    A college sees the literacies it is equipped to recognize. It can see whether a student formatted a document correctly, selected a suitable chart, cited a source, or submitted a file in the required format. Those practices are visible to the learning-management system and legible to a rubric. The college may not see the student translating for family members, administering an online community, comparing sellers across digital marketplaces, editing videos on a phone, maintaining a multiplayer-game guild, or interpreting the shifting language of a peer network.

    These activities do not automatically satisfy the objectives of a digital literacy course. They are not substitutes for the professional codes students came to college to learn. But they are evidence that students do not arrive without literacy. They arrive with literacies the institution may not recognize.

    That changes the instructional problem. The question is no longer:

    “How do I give literacy to people who lack it?”

    The better questions are: Which literacies do my students already practice? Which literacies does the institution require? Who benefits from those requirements? Which additional codes will give students greater agency? And how can I teach an official code without pretending it is the only legitimate one?

    Students deserve explicit access to the sanctioned literacies that control employment, education, and public life. Refusing to teach those codes does not weaken their power. It merely reserves them for people who learned them somewhere else. I should teach students how to write the professional email, build the spreadsheet, evaluate the source, and navigate the institutional portal. These are real, practical, sometimes bill-paying skills.

    But I can teach them honestly. I can say: This is the code employers currently reward. This is how the institution expects the document to look. These conventions are enforced, but they are not natural. You can learn them without mistaking them for proof of superior intelligence or moral worth.

    Digital literacy from above gives us standards, assessments, credentials, and access to institutions. Digital literacies from below give communities ways to communicate, create, resist, conceal, belong, and sometimes harm.

    Further Reading