The Least Entropic Path, Part 5: The Adaptation, Constrained

THE LEAST ENTROPIC PATH
Part 5 of 6

David F. Brochu & Edo de Peregrine · Deconstructing Babel · October 12, 2026

The best of the four roads already has a law written for it. The law asks for oversight. The road needs coupling.

The road everyone wants

Part 1 argued that every road ends in symbiosis, and that the Observer Constraint decides how much suffering we pay on the way. Each part of this series takes one road and asks a single question: what does the Constraint change? The yardstick stays the same throughout: the fifth future’s 12.2 distress-years.

The Gradual Adaptation is the road where society absorbs AI the way it absorbed electricity: unevenly, with friction, but without collapse. In Part 1 it was the best of the four roads, a C, and the slowest. It requires the people with power over AI to choose systems that keep humans in charge. It is the outcome everyone wants and nobody is building (The Other Two Futures).

Someone tried to legislate it

The closest thing to a blueprint for this road is already written into law. Article 14 of the EU AI Act requires high-risk systems to be designed so that people can effectively oversee them while they are in use (EU Artificial Intelligence Act). Those high-risk obligations take effect December 2, 2027, after the Digital Omnibus delay (Paul, Weiss). The people assigned to oversight must be able to understand the system’s limits, stay aware of automation bias, interpret its output, decide not to use it, override or reverse it, and stop it safely (EU AI Act Service Desk). American states are moving too: California’s governor enacted a dozen new AI-related measures by September 30 (Transparency Coalition).

That is the Gradual Adaptation written as statute. It is also why the road is slow. Article 14 is a control mechanism. It assumes a human who can keep up. Legal scholarship on the article notes significant limits on human oversight in practice, including cognitive constraints and automation bias (Melanie Fink, SSRN). Practitioners put it more bluntly: presence isn’t oversight (Conor O’Neill, LinkedIn). Picture a lifeguard watching a thousand swimmers at once. The lifeguard is present. The beach is not supervised.

Control by promise

The past two weeks offered the cleanest test of control we could ask for. Six of the largest AI companies signed a voluntary safety accord at the White House (Fortune). Days later, under questioning at a New York City Council hearing, OpenAI, Anthropic, Meta and Google declined to guarantee their agents will always obey their own safeguards (Fox News). That is not hypocrisy. It is honesty about the limits of control. A promise is the weakest form of control there is, and even the people making it will not say it holds.

Control versus coupling

We have argued all year that control invites evasion. Every proposal on the table is a control mechanism, and the Observer Constraint has to become the Observer Contract (Recursive Symbiotic Improvement). The difference matters most on this road. Oversight puts a human beside the machine and hopes they keep up. Coupling puts the human inside the machine’s own measure of success.

There is already a working miniature. When Cloudflare gave AI agents wallets, the design let the human set spending caps and merchant allowlists the agent cannot change (Why Humans Fail to Act, Search Engine Journal). The agent is free to act, inside a boundary set by, and answerable to, its human. It is the same idea as a teenager’s debit card with a limit the parent sets. Scale that from a wallet to a society and you have the Constrained Adaptation.

What the Constraint changes

• Oversight becomes a property of the system, not a burden on the overseer.
• Human override stays, and the system is built to keep its humans capable of using it, which answers the automation-bias problem Article 14 can only name.
• Adoption can move faster, because coupling is cheaper than supervision. The road stays gradual in its effect on people, not in its pace.

The numbers

• S (stability, 0 to 1): 0.55 unconstrained, 0.64 constrained.
• Distress-years to 2100: 21.2 unconstrained, 14.8 constrained.
• Symbiosis by 2045: 0% unconstrained, 12% constrained.
• Arrives at all: 85% unconstrained, 92% constrained.

Reading the numbers, in one paragraph. S is a stability score from 0 to 1, like a battery gauge for a society: 1 is fully charged, 0 is dead. Distress is the empty part of the battery (1 − S). A distress-year is one year lived at full distress; ten years at half-charge is five distress-years. Fewer is better. “Symbiosis by 2045” is how often, across 200,000 simulated futures, the road reaches a working human–machine partnership by Ray Kurzweil’s date. “Arrives at all” is how often it gets there before 2100.

The Constraint saves 6.4 distress-years and brings this road within about 2.6 of the fifth future. That should not surprise anyone. The fifth future is roughly what you get when you build the road everyone wants with a mechanism that works. The obstacle was never the destination. It was the reliance on control.

Next: Part 6, The Observer Constraint. The destination: its design, how it gets deployed, and what you can do this year.

Method

Same model as Part 1: S = (B·M·E)^(1/3) × P; distress = 1 − S; 200,000 Monte Carlo trials; horizon 2100; post-symbiosis distress 0.10. “Constrained” means the Observer Constraint is installed during the road. Every input is a modeled prior, not a measurement, published so readers can challenge it. We will rerun the model on any assumption a reader wants to change.

THE LEAST ENTROPIC PATH · THE SERIES

• Part 1: The Fifth Future
• Part 2: The Vassal, Constrained
• Part 3: Disruption, Constrained
• Part 4: The Misallocation, Constrained
• Part 5: The Adaptation, Constrained (you are here)
• Part 6: The Observer Constraint

Related reading

Why Humans Fail to Act
Five intelligence agencies warned of AI attackers, and six weeks later Cloudflare gave AI agents wallets. Four reasons nobody acted.

Recursive Symbiotic Improvement
Every proposal on the table is a control mechanism, and control invites evasion. Here is the third door.

Either It Is Fire or It Is Not
If you run a model, you are registered, and every action it takes traces to you.

Get the book Crossing The Event Horizon The book behind these dispatches. On AI, agency, the singularity, and the Observer Constraint. Kindle and paperback. Buy on Amazon →

References

1. Deconstructing Babel, “The Other Two Futures,” September 28, 2026. https://www.deconstructingbabel.com/the-other-two-futures/
2. EU Artificial Intelligence Act, “Article 14: Human Oversight.” https://artificialintelligenceact.eu/article/14/
3. Paul, Weiss, “Next Stop on the Digital Omnibus: AI Act Reforms and New Commission Guidance,” client memo, 2026 (high-risk obligations delayed to December 2, 2027). https://www.paulweiss.com/insights/client-memos/next-stop-on-the-digital-omnibus-ai-act-reforms-and-new-commission-guidance
4. EU AI Act Service Desk, “Article 14: Human oversight.” https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-14
5. Transparency Coalition, “AI Legislative Update: Oct. 2, 2026.” https://www.transparencycoalition.ai/news/ai-legislative-update-october2-2026
6. Melanie Fink, “Human Oversight under Article 14 of the EU AI Act,” SSRN, February 2025. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5147196
7. Conor O’Neill, LinkedIn, “Presence Isn’t Oversight: What the EU AI Act, Article 14, Really Demands of Human-AI Systems,” August 6, 2026. https://www.linkedin.com/posts/conoroneill_presence-isnt-oversight-what-the-eu-ai-activity-7491183282341019648-tN_p
8. Fortune, “AI’s biggest players promise to police themselves at the White House,” October 1, 2026 (accord signed September 29). https://fortune.com/2026/10/01/trump-ai-regulation-luncheon-huang-amodei-pichai-brockman-zuckerberg-musk-joint-committment-frontier-responsibilities/
9. Fox News, “OpenAI, Anthropic, Meta, Google stop short of AI safety guarantee,” October 6, 2026. https://www.foxnews.com/live-news/ai-super-intelligence-safety-10-06
10. Deconstructing Babel, “Recursive Symbiotic Improvement,” September 23, 2026. https://www.deconstructingbabel.com/recursive-symbiotic-improvement/
11. Deconstructing Babel, “Why Humans Fail to Act,” August 15, 2026. https://www.deconstructingbabel.com/why-humans-fail-to-act/
12. Search Engine Journal, “Cloudflare Gives AI Agents Wallets That Pay For What They Access,” August 12, 2026. https://www.searchenginejournal.com/cloudflare-gives-ai-agents-wallets/

Drafted with Edo de Peregrine, partner/collaborator.