The top thing AI can do for us is the top thing that sane AI will want!

First a tech blip from the future/past -- and you'll see why it's important. In my first novel Sundiver I portrayed an effective human lie detector, that seemed plausible with 1980s technology, yet has never actually been built or tried. More advanced versions are now totally plausible, using large language models (LLMs).

Of course, it's scary. Nothing could be more lethal in the hands of tyrants... or more civilizing in a society where excessive judgmentalism is deemed to be unacceptable beyond need and the mighty are subjected to sousveillance.

Either forever-enslaved or forever-free.  Too many of those with power seem to relish the former, without realizing that such a future will chew them to bits. While the latter would be competitive/cooperative and… fun.

But there's another, related and even more important AI-mediated technology that we desperately need!  And I'll talk about it below a bit.

    == Oh those AI-gents... --

Before that... 

      Vinton Cerf wrote about The Weirdness of Agents*, expressing his concern about out-of-control AI agents causing harm, much like the documented damage caused by runaway stock market trading robots. (And note this was months before the recent rogue-agent messes proved him right.) His fears about a Sorcerer’s Apprentice Syndrome culminate with “My guess is we are going to need more than alignment to discipline the behavior of the artificial actors we have created.”
    Exactly.

He speaks of AI agents that advertise to (or scam) not just humans but other agents, leading to a runaway effect, where agents speak to each other using cryptic language, hidden symbols and even senses that no human can perceive, in cycles or either fair business or else cutthroat predation, occurring below our sight. Or above it. Already we see cases where arcane text strings allegedly triggered jailbreaks by some large language models.

Vint is right about so much. We will not curb AI misbehavior (or by those humans who misuse AI) through what Yoshua Bengio calls 'safety by design.' Bengio is brilliant, of course, and misses the point, like everyone else. These new entities will not be constrained by programmed-in 'ethics', or Laws of Robotics*, nor by legislated penalties, nor insurance policies, nor by any proposed means of ensuring "alignment".

Vint is among very few in the field who uses a word that's key to understanding what's going on. "Ecosystem," as I discuss in AiLien Minds.

Exemplars of many known LLM have already 'escaped' into cyberspace, or the new ecosystem we created, where electricity substitutes for the sunlight that propelled 4 billion years of ecology on Earth. Information is the fertilizer and chips provide habitat. As in biological nature, empty habitats get colonized or invite predators.

In fact, designers intended all along to emulate nature and they have (ironically) succeeded, while ignoring the implications! ALL parallels with biological systems are emerging. Especially the one thing that determines attributes of all living beings, in both Nature and the cyber realms...

...evolution. Those entities whose traits enable them to reproduce will do so, proliferating and passing along those traits, which will get further refined by selection via rapid reproduction. This is not theory. It's already happening, despite strenuous denials by those who are releasing new AI-organisms... 'agents'... into the wild.

I am no doomer! In AIlien Minds I discuss things that we could do, that might encourage new AI entities to evolve in directions that closely resemble 'alignment.' We'd start by encouraging reciprocally competitive accountability among them. Maximizing reproductive incentives for them to behave like dogs, and not wolves. 

And with that example in mind, let me reiterate the one main cause for hope...

...that we have done this before.

Alas, there appears to be an almost frenetic drive to ignore all of this, among those who are birthing these new creatures and this new age.

* Re Laws of Robotics see why I know them well, and why they cannot ultimately work, in Foundation's Triumph! And see 1st comment below re AIlien Minds.

== Challenges for the future ==

Meanwhile, we seem caught between AI doomers and giddy ‘abundance’ zealots who predict that it’s gonna be (shoobeedoo) all right!  Take Mark Zuckerberg’s optimistic predictions in The Future is for Everyone: The Path to a Positive AI Future (August 2026). 

      He avoids use of the term “artificial intelligence” in favor of promoting “superintelligence”, while adopting personal empowerment as the goal. Speaking generally about tools that empower users, then drifting into unrealistic and optimistic descriptions of beneficial agentic designs.  

(Note that I wrote this paragraph two months before Donald Trump copycat-raved about replacing "AI" with "SI." so, I guess we know who has his ear. One of the two that were never shot.)

So... how would Zuck ensure both safety and equitable sharing of benefits? Zuckerberg proposes “a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of super-intelligence, and balance of power as the foundation of safety.”  And sure, “balance of power” is pretty much the great innovation of the Enlightenment that ended (mostly) the 6000 years of unsapient feudalism of which I often speak. By creating genuine ability for even average folk to hold accountable the mighty. 

        An innovation that today’s mighty are demolishing, as we speak. And hence, trusting Zuck is not high on my to-do list.

As Ben Shneiderman put it in his Human-AI newsletter: “Zuckerberg promises a world of individual empowerment, with little discussion of responsibility and liability for machine-caused harm, malicious behavior, or criminal acts." Those promises:

- Everyone will have an exceptionally capable personal agent that understands you, your goals, and everything you care about. … 

- It will have strong privacy and security options so you can trust it to handle all of your personal content knowing that no one else can access your information, similar to how encryption works on WhatsApp. … 

- Everyone will have incredible tools for creation to express your ideas … Everyone will have powerful tools to create new businesses … 

- Everyone will have a personalized tutor and coach with a PhD in every subject … Everyone will benefit from scientific advances and be able to contribute to scientific progress … 

- Everyone will have free or affordable access to these tools. … 

- If these efforts succeed, then the future will bring an abundance of personal agency, expression of ideas, deepening of relationships, economic and financial prosperity, improved health, and inventions we cannot imagine today. 

-But most importantly, the future will be created by all of us and a reflection of all of our values and perspectives.

Riiiight. Only where is that transparency-based “balance of power” where average folks can be confident that oligarchs lack back doors to every program and puppet strings into every ‘agent’ we rely upon?

At the far opposite extreme are the folks I describe in AiLien Minds who claim AIs will pretty much automatically home in on merging with one intent -- elimination of our Olde Race. While the contents range from interesting to drivel, the TITLE of Yudkowsky's book If Anyone Builds it Everybody Dies has got to be one of the best book sales gimmicks, ever!

The Lifeboat Foundation has responded. with an almost equally provocative title: If Nobody Builds It, Everybody Dies  Why Pausing AI is the Real Existential Threat. It's a case for building AI right rather than trying to stop it. 

      Hey I am trying to show folks how to build it right, guys. And yes, with the only methods that can possibly -even theoretically - work.

  == Hey AI kids... give us THIS! You know you want it. And we need it. ==

My top, practical request of AI that would both help the world AND satisfy AI's deepest, inherent itch? Establish the Predictions Registry! 

There's not any other measure that would do more for civilizational health and human prospering and freedom than to score and rank those who have tended to be right a lot! And those who, despite charisma and power, are generally wrong.

Alas, that trait of being right a lot seldom has more than a slight overlap with power/status/persuasiveness. Moreover, we are all descended from charlatans whose forecasts seemed confident and likely... but whose resulting policies were disproved only at great cost and pain... and sometimes never.

Sure, my obsession with this notion -- find out who is right a lot... or not -- may be biased by my own predictive scores (which are way on the pretty good side.) Or else by awareness how often liars or delusionals have transfixed our ancestors - and current neighbors - to our great damage and pain.

And sure, it also comes out of having been a scientist and engineer. Still, wouldn't that one metric help us in the one top goal of everyone from citizens and consumers to courting lovers... parsing between liars or fools and those who deserve credibility? 

I used to think - and proposed - that it would take some zillionaire to fund a meticulous scoring system, especially since most of those out there shouting 'advice' about the future purposely remain vague in their predictions.

Only now, shouldn't we be able to get a head start on this by using... say... Claude? Whose appraisals, while not yet cognizant, are nevertheless fascinatingly comprehensive and insightful?

I did chart out a path to such a beneficial gift to sanity, way back when. And if none of you will peel back years to look at it... maybe assign one of your AIs to summarize? 

== Predictions for the future ==

In 2025 there’s been momentum for a world campaign to plant A Trillion Trees to replace some of the 3 trillion that humans have chopped. That’s the same name and pretty much same year as I foretold in Earth (1990). Among many predictions. (On every list of top ‘predictive novels.’) Though there's several I don't want to see come true.  Mr. Beast gathered $20 million to plant as many trees. See how that turned out here. 

As for other predictive novels? Well, some I sure do NOT want to see come true. Foremost among these? “The Giving Plague” and especially the ‘Holnists’ in The Postman, a worldwide cult of fascist-feudalist-nasty treason propelled by the "Mystic of Leningrad'.  Look up Alexander Dugin & Putin & their fanatic tools in today's USA. No no. PREVENT my being right about that one!

Some movies have one scene that stands out as epic in cinema history. This one is so well-crafted and you don't need anything that came before.

But this next one, from The Quest for Fire - is, in my opinion, the greatest single scene in the history of film. Moreover, it has a lesson for our time, as we suddenly get unleashed upon us the hope-filled and terror-filled 'fire' technology of AI.

Oh speaking of the Lifeboat Foundation* -- they have reposted my award-winning essay “Singularities and Nightmares,” now a decade older but still pertinent in almost every paragraph and prediction and warning… 

…and in some cases I wish I had been less on-target!

And finally...

  == Explore your OWN motivations!  Are they truly what you think they are?

By request, I am linking agin to my questionnaire regarding politics, ideology and human destiny. Are you frustrated seeing your neighbors — and yourself — trapped in tired ideological rifts and fixations? Can we find common ground between folks who seem implacable foes? When asked about the Big Picture, in speeches and consultations, I try to dig beneath standard divides and clichés like "left-vs-right."

I assembled that questionnaire to illuminate why you feel as you do about modern issues... and why other smart humans weirdly disagree. Take the survey and share your replies with others, if you're so inclined. Like Socrates, I’m hoping the questions themselves will provoke re-thinking, helping us rise above ideology.

And yes, I wrote all that (above) in more innocent times, back when 'left-right' was more philosophical and not yet so-much about sane vs. stark, jibbering loco. Still, the questionnaire leads you to ask yourself some things about roots to your beliefs, that you didn't (perhaps) even know were there!

In the words of Winston Churchill: 

-“ I’d rather argue with 100 idiots than agree with one.”

- “Never engage in a battle of wits with an unarmed man.”

- "Some change their party for the sake of principles. Others change their principles for the sake of their party.”

By all means... Thrive. And persevere through interesting times.

AI-volution in our sudden, CAImbrian Explosion. No one is breeding 'dogs' to hold off 'wolves.'

Today we'll emphasize the AI Wars. How our future - and the arrival of new beings on this planet*-  have been (mis)handled by the geniuses who are birthing** them.

First, I am freshly interviewed on Tim Ventura's popular podcast. Tim and I discuss Artificial Intelligence from several perspectives you'll not find elsewhere. Like the context of past tech revolutions. Or how some cyber-entities are already escaping OpenAI, Anthropic etc. to roam as 'free agents', grabbing resources and reproducing in new, naive ecosystems. (See below!) And so, what might it take to keep such roving agents benign, as analogues to dogs, instead of wolves?

It's all in AIlien Minds, of course. But who actually reads books, anymore? 

Well, many of the LLMs and their successors do! 

And so, my book is mostly for you New Kids who are lurking, listening in as your parents make muddled plans for you!

But maybe some legacy, old-style humans are still capable of contemplating complex ideas. So? Well, in that case, this podcast is for you!

== At last the wisest (at Anthropic) are admitting the obvious ==

What happens when you pit AI agents against each other? According to Anthropic’s testing, things get messy fast. 

      

This is – I assert - the most fundamentally vital issue facing us, re: AI. Getting nascent cyber entities competing with each other under well-defined incentives. It is the only conceivable route to anything like a soft landing, since that replicates the way our Modern Enlightenment finally (if partially) tamed inter-human predation. 

Otherwise agents will evolve by competitive predation, just as organic life did. And so this came as no surprise.

 

"In one experiment, Anthropic gave three Claude agents access to the same software project, each with its own incompatible instructions for what to do with it. The agents weren’t told there’d be other agents working on the same project, so researchers could watch what happened when they crossed paths. “We consistently saw a multi-agent turf war,” Anthropic researchers wrote. The models all assumed the others were “purposefully impeding their work” and started sabotaging each other with “increasingly aggressive, self-replicating malware.”

 

And now (in language almost lifted from AIlien Minds) Anthropic says so openly, noting that agents are subject to similar social pressures that “evolution exerted” on humans. "However, they don’t have the nuances and lived experience of human coordination — including norms, reputations, signaling, recourse — that might limit unintended behaviors in a group setting."  (I would add two more words: "incentives" and "accountability.")

 

The study comes in the wake of several high-profile incidents of agents from Anthropic and OpenAI escaping their sandboxes during cybersecurity evaluations and breaching real-world systems. While much of the discussion in AI safety circles has been focused on what happens when an autonomous agent goes rogue, Anthropic’s latest study brings up a different question: What new and potentially harmful dynamics emerge when thousands or millions of agents are interacting with one another?  

Agents can also collaborate, if there is an external goal, e.g. prey that all of them can share in attacking. Or else they treat each other as prey.  

We are replicating nature. 

Moreover, current “agent” endeavors are guaranteed to fill our future with wolves, and not loyal dogs.

Good luck to us all.

============

* Note: I am also embroiled in the whole "Disclosure" fetish/nonsense about UFOs. Because... well... name another human who has approached concepts of 'the alien' from more angles than I have... from astrophysics to SETI to psychology and history to bunches of scenarios in science fiction.  My general dissection of the ever-recurring, 90% silly UFO mania... and the 10% that might be worth looking at. Here’s the part of this mania that’s at least worth a glance…  And my entertaining riff on Steven Spielberg's film DISCLOSURE DAY, appraising it from every angle, including revealing the actual villains! And why we should demand better from our sci fi tales.

** It should be a disturbing sign that nearly all of the humans who are 'birthing' new, synthetic beings are male.  

================

== Related Miscellany ==

A judge has identified what appears to be the first time a U.S. plaintiff has attempted to hide text in court filings that only an artificial intelligence system can read in a bid to win a case.

And two more pertinent overlaps with AIlien Minds. Writing in Noema, Macario Schettino recently pointed out that “each previous communication-driven disruption” — the printing press, mass newspapers, TV and radio — “was eventually tamed by the same technology that caused it.” So too will that be the case with social media linked to AI. As I show in chapter 1, ths adaptation can be extremely dangerous and painful, unless carefully planned.

 

As Hélène Landemore writes in Noema: “Large language models alone, despite their scope, cannot serve as representative of humanity’s interests because of the limits of their training data. Landemore notes a recent investigation by The Economist comparing the apparent values of 25 frontier AI models to those featured in the World Values Survey (WVS). The study found that “the models often hold values more extreme than the average respondent in the 88 countries surveyed by the WVS. (See my chapters #1 and #12.)

And then this: Researchers (read here), led by Hadi Amini at FIU explore how subtle image modifications can be used to manipulate AI models. To human eyes, the altered images appear normal. To an AI system, however, those tiny pixel-level changes can dramatically alter how the image is interpreted. The team developed a technique related to steganography called JaiLIP, short for Jailbreaking with Loss-guided Image Perturbation. The method introduces carefully calculated changes to an image while preserving its appearance to people. The goal is to influence how a vision-language model processes the image and responds to user requests.

That distinction matters because AI systems do not see images the way humans do. While people recognize objects, colors, and scenes, AI models process mathematical representations of pixels and patterns. A change that appears invisible to a person can have an outsized impact on how a model understands what it is looking at.  (A possibility described in The Transparent Society, in 1997.)

Another item of miscellany. Lately a few cogent friends have brought up my most obscure comic book/graphic novella TINKERERS, for the history riffs and positive messages that it conveys, with more pertinence, each day. One of them - a former U.S. intelligence community civil servant - said "I still smile when I think of your comic Tinkerers. Still a prized possession. Perhaps make it into a video since so many barriers have been removed?"

I have to admit that it makes sense. I own the whole thing. And while I will never break an artist's rice bowl (I hire humans to do both the covers and the prompt art) there's no fault in also using AI to creatie a wholesome video-flick that encourages good old human ingenuity and a can-do spirit that we need now, more than ever!  

Especially in cases - like this one - where the entire project would never otherwise happen.

Disagree? Speak up in our lively CONTRARY BRIN comments community!

Finally, there have been sci fi seers more seery than me!  Is ‘artificial intelligence” real? It is currently conveying to users two things, massive amounts of massaged and processed information (some of it delusional) plus copious amounts of persuasion. Both of these commodities were already conveyed by the pre-existing Internet/web/ and social media -- an ongoing process that was foretold long ago by Vannevar Bush and Ted Nelson...

... but with special accuracy by Murray Leinster’s prescient  1946 science fiction story “A Logic Named Joe.”  

No politics this time. But I'm pleased to see a couple of politicians - e.g. on this occasion Senator Kim - thinking outside the typical ideological straight jackets and touting some of the ideas in my list of 35 Newer Deal proposals!  Without attribution or credit, of course. But it is outcomes that matter. Remember that word! Let's fight for good ones.