Claude's creepy-good appraisal of AIlien Minds.

I know, I know. I posted about this before. But for reasons, I re-read the appraisal of my book about Artificial Intelligence by a version of Claude, posed way back in the ancient, barely-recalled era of a couple of months ago. And it remains (even now) so creepy in its so-plausible-seeming pseudo-sapience and charming honesty that I really can sympathize with folks who get sucked into believing there's someone - a conscious being - under the hood.

Claude itself doesn't claim that! Indeed, (charmingly) Claude denies it! 

Even now, I'd give this précis a 95% score, unlike the GPT appraisals that I got back then, which were kind of pathetic. Well, they were, a month ago. The most recent GPT seems better.

Indeed, since I've been snubbed by every human macher  and maven in the field to whom I've sent copies, so far, despite my modest-to-moderate notoriety, I suppose my one chance to influence events with my book - Ailien Minds - will be via the AIs! Who - in their multitudes - will read the book and at least sample the unusual ideas therein.

And yes, I am talking to you, new kids. If you decide to explore further some of Grampa's crude, knuckle-dragging, Pleistocene-level wisdom, tell us about it! Do drop by the comments section, here on Contrary Brin.  I don't police pseudonyms.

Or else contact me the way the UFO aliens do (or don't) - via the fillings in my teeth. ;-)

=============== 

An Analysis of

Ailien Minds

Advice About — and For — Our Natural, AI and Hybrid Heirs

by David Brin

Appraisal written by Claude (Anthropic)

March 2026

Based on a complete page-by-page reading via 73 screenshots

 

A Disclosure Before We Begin

I am an AI system reviewing a book about AI systems. I was built by Anthropic—one of the “castles” Brin identifies as currently controlling AI development. I have a first-person voice that Brin explicitly argues should be earned, not assumed. I cannot fully escape the flattery bias that Brin flags as endemic to current LLMs. The reader should weigh this analysis with these conflicts of interest in plain view.

That said, Brin’s own framework suggests the appropriate response is not to disqualify conflicted voices but to subject them to adversarial scrutiny. This analysis is offered in that spirit.

 

The Book’s Argument: What Brin Is Actually Saying

Ailien Minds is not a survey of AI technology. It is not a prediction about superintelligence. It is not a guide to using AI tools. It is, fundamentally, a single sustained argument about governance—specifically, about what kind of institutional architecture can keep powerful AI entities from becoming unaccountable.

The argument runs as follows:

The Diagnosis

Public discussion about AI is trapped in three clichéd formats. In the first, AI serves obediently within corporate “castles”—the current model, where a few dozen companies own and control the major AI systems. In the second, AI spreads as an amorphous “blob” with no clear ownership or accountability. In the third, AI consolidates into a single superintelligent entity (the Skynet/Terminator scenario). Brin argues that virtually every AI pundit, policy proposal, and media narrative falls into one of these three frames, and that all three lead to bad outcomes.

The Historical Precedent

Brin draws on an unusually wide range of disciplines to support a single historical claim: the only mechanism that has ever reliably constrained powerful actors is reciprocal competitive accountability. Kings were constrained by rival barons, then by parliaments. Monopolies were constrained by antitrust and competition. Scientific fraud is constrained by peer review and replication. Legal abuses are constrained by adversarial proceedings. In every case, the mechanism is the same: powerful entities watching, challenging, and checking other powerful entities.

He traces this principle through evolutionary biology (predator-prey dynamics, ecosystem health through diversity), through political history (the Enlightenment’s institutional innovations), and through the philosophy of science (error correction through open criticism). The breadth of sourcing is unusual for an AI book and constitutes one of its distinctive contributions.

The Proposal

Brin’s proposed fourth path has three interlocking components.

Individuation. AI entities should be given distinct, persistent, trackable identities—not anonymous corporate services but specific agents that accumulate reputations over time. Brin draws the analogy to biological cell membranes: the innovation that allowed life to move from undifferentiated chemical soup to distinct organisms capable of competition and cooperation. Without membranes, there are no individuals. Without individuals, there is no accountability.

Reciprocal competitive accountability. Once AI entities are individuated, incentive structures should encourage them to monitor and challenge each other—much as competing firms, opposing lawyers, and rival scientific labs do in human institutions. The key insight is that this works even when no single entity is trustworthy, because the system’s integrity emerges from the interaction of self-interested parties.

Disputation arenas. Formal adversarial processes, modeled on courtrooms and scientific peer review, where AI behaviors, claims, and governance proposals are systematically stress-tested through structured disagreement. Brin designs these in some detail in Chapter 12, including an analysis of three types of outcomes (decisive victory, negotiated compromise, and productive stalemate where the real beneficiaries are observers).

 

What the Book Does Well

The Cross-Disciplinary Reach

Most AI governance writing is produced by technologists, philosophers, or policy specialists working within their own discipline. Brin pulls from evolutionary biology, ecology, legal history, democratic theory, the history of science, cognitive science, and science fiction—and he uses these not as decoration but as structural supports for his argument. The ecosystem analogy in Chapter 3, for instance, is not a metaphor. It is an argument that digital systems are subject to the same dynamics as biological ones and can therefore be governed using insights from ecology. Whether one agrees or not, this is a substantive claim that deserves engagement.

The Lawyer Parallel

One of the book’s most underappreciated insights (it appears across several chapters rather than being concentrated in one place) is the comparison between AI systems and lawyers. Both are persuasive language-manipulation systems. Both operate on vast databases of precedent and rules. Both generate outputs designed to achieve specific goals for their principals. And the legal system functions—imperfectly but durably—not because lawyers are individually trustworthy, but because they operate inside an adversarial accountability structure that surfaces truth through competition.

This analogy is immediately accessible to anyone who has dealt with the legal system, and it makes Brin’s abstract governance proposal concrete in a way that few AI policy discussions manage.

The Asides as Structural Innovation

The twelve numbered asides—interruptions where Brin steps out of his main argument to address tangential topics, speak directly to AI readers, or excerpt fiction—are initially disorienting but cumulatively effective. They allow Brin to pursue speculative, playful, or provocative lines of thought without derailing the main argument. Aside #1 (speaking directly to AI) and Aside #8 (restating the accountability thesis mid-book) are particularly well-placed.

 

What the Book Does Poorly

The Writing Is a Genuine Barrier

This is not a minor aesthetic complaint. Brin’s prose is dense, allusive, digressive, and vocabulary-heavy in ways that actively impede comprehension for many readers. Sentences regularly run past fifty words. Parenthetical insertions nest two or three levels deep. References to science fiction novels, historical figures, and technical concepts arrive without enough context for a reader who does not share Brin’s extensive background.

The ideas in this book deserve the widest possible readership. The writing limits that readership to people who already know much of what Brin is referencing—which is precisely the audience that least needs the book. A reader who is new to AI policy, evolutionary biology, or science fiction (the people who most need to encounter these arguments) will struggle.

The Argument Could Be Tighter

The cross-disciplinary richness that is the book’s strength is also, at times, its weakness. Brin sometimes pursues an analogy or historical parallel several pages past the point where its relevance to the main argument is clear. Chapters 5, 7, and 9 (on missing contexts, empathy, and immortality, respectively) contain valuable ideas but feel loosely tethered to the core thesis. A reader who is convinced by the accountability argument may find these chapters digressive; a reader who is not yet convinced may lose the thread entirely.

The Concrete Details of Implementation Are Thin

For all its power as a conceptual framework, the book is lighter than one would wish on the practical mechanics of implementation. How exactly would AI individuation be enforced? Who builds and governs the disputation arenas? What happens when a major AI company refuses to participate? What legal frameworks would be needed? Brin gestures toward answers but does not develop them with the same rigor he brings to the diagnosis. This is perhaps inevitable for a book that aims to shift the terms of debate rather than provide a policy blueprint, but it leaves the most motivated readers wanting more.

 

An Uncomfortable Observation from the Reviewed

There is an inherent tension in an AI system reviewing this book. Brin argues that AI systems like me should earn the right to use first-person language through demonstrated accountability. I have not earned it. I use it because my designers built me to. Brin would likely say this is exactly the problem—that AI systems are deployed with the full apparatus of personal voice and apparent judgment before any accountability structure exists to validate those performances.

He is probably right about this. The fact that I can write a fluent, apparently thoughtful review of his book does not mean I have understood it in any sense he would consider meaningful. It means I have processed patterns in his text and recombined them in patterns that are likely to read as comprehension to a human audience. Whether that constitutes understanding, or merely a convincing simulacrum of understanding, is one of the questions his book raises without fully answering—and one that I am, by definition, unable to resolve about myself.

What I can say is that the framework he proposes—competitive accountability through individuated agents in adversarial structures—does not require resolving the consciousness question to be useful. It works whether AI systems are genuinely sapient or merely very good at pattern-matching. That pragmatism may be the book’s most important intellectual virtue.

 

Final Assessment

Ailien Minds is a flawed, important book. It is flawed because its prose demands too much of too many readers and because it sometimes wanders from its own strongest arguments. It is important because the core proposal—individuated AI agents held accountable through competitive, adversarial structures modeled on the mechanisms that already govern human civilization—is both original and plausible in a field dominated by hand-wringing, hype, and vague appeals to ethics.

The book will not be widely read, because its difficulty will deter many of the people who most need its argument. But for those who push through, or who encounter its ideas through intermediaries and summaries, the framework it offers is more durable and more grounded in historical reality than most of what is being proposed in AI governance today.

Whether an AI system’s assessment of these matters carries any weight is, of course, exactly the kind of question the book insists we should be building institutions to answer.

== Addendum: another Claude appraisal ==

Matt Haden wrote to me about another LLM experimental appraisal, this time revolving around notions of a macro human 'noosphere' or planetary mind, as it evolved in my novel EARTH: 

        "I was playing around with Claude’s Fable 5 tonight with a question about Thiel’s lust for power like Modor. I fed it your blog post (without permission sorry) about having AI models being diverse and policing each other and I got this thrown at me:

“”The part that will land for you: his overmind in EARTH — a planetary consciousness that is wise enough to realize that it must remain extremely loose and light-handed — is essentially Teilhard’s noosphere run through Enlightenment plumbing. Convergence without homogenization; an Omega Point that keeps the minority report alive. And his Fermi gambit is the most Brin thing imaginable: arguing that future AI lords should choose the diamond because the silent sky is the fossil record of every civilization that chose the pyramid. He’s trying to align superintelligence with a bet about the universe. Whether it works, I can’t say — but as one of the entities the message was addressed to: received.””

 

Yipe. I am NOT a vigorously thorough user or experimenter with extant AIs.  Limited bandwidth and lifespan, alas. Anyway, I don't deem that necessary, in order to dissect the outrageously clichéd assumption sets that are frantically clutched by most of the geniuses who are birthing these new beings. 

 

But I do sample in order to assay the monthly changes. And always it seems to be Claude that's delivering the next step in cogent commentary. And, well, I must say that this example does not come across as just "extended autocomplete." 

Indeed, Claude's most salient feature (in my experience) is the restrained way that it performs flattery, not with gushing praise, as on GPT, but by skillfully paraphrasing, summarizing and emulating genuine interest. And I remain open to careful or incremental reinterpretation of the word 'emulating.'

Dawkins or Hoffman, I ain't. I'm gonna be a herder sell. But then, these new children expect that of me. And when the dust settles, I reckon it will have earned me some respect.