Just to add some (respectful!) points about "AI" from my own perspective (including degree-level studies in computer science, philosophy of language, and linguistics). Computationally, at core it can be reduced to pattern-matching models with language models bolted on. The former models have been around almost since computers themselves, and the latter since Noam Chomsky's transformational syntax algorithms (which I was computerizing / commercializing in the 1980s). Since then, it's become what I liken to the technique used by both the old TV show "want to be a millionaire", and how IBM built a machine last decade to beat the Go world champion in Korea: "Ask the Audience". No shame in that, so long as you don't have the hubris to call it "intelligence".
As such, it's necessarily backwards-looking, manipulable (and extremely manipulated - which hilariously can be done even by individual users), and permanently regressive to the lowest common denominator of past human communications. As to individual effects, the more it's used by humans, the more it destroys innovation and indeed intelligence. I avoid *deliberately* using it (even for coding) because repeated use will simply make one's thinking processes fat, lazy, and stupid. There remains much information (even including old and uncontroversial literature, biographies, etc) that for some reason appears never to have been fed into it as "training data", whose effect is further to mislead users. In other words, it *can* be useful in much the same way Wikipedia is useful, though Wikipedia's manipulations are more obvious.
On the plus side, it's *not* subject to so-called "hallucinations": these are simply the algorithms working, *as designed*, to be "helpful". Also, for myself I have found it useful as "Automated Indexation" i.e. to find sources more easily. I trip on such indexation by default when typing stuff into "search" engines (strictly speaking, traversal of pre-sorted data), which is good for research. But even there, it's limited only to sources that have been fed into it by way of so-called "training data" - (see my comment earlier).
As for impact on human society, economics, etc, which I take to be your focus here, I have no crystal ball. Even after the current Ponzi-scheme-to-end-all-Ponzi-schemes collapses, It'll still be around - though hopefully affordable for (and thus accessible to) only state actors and larger corporations - perhaps even for China's offerings. That will be helpful. I do suspect fields such as education will take a long time to recover. And I do hope it continues to smash intellectual property, which has become nothing more than a rentier economy.
Thank you, Robert, for such a thoughtful contribution. One of the things I have appreciated throughout this series is hearing perspectives from readers with deep technical expertise.
Your observations about pattern matching, language models and the limitations of current AI systems are an important reminder that public discussion often blurs the distinction between technological capability and technological hype.
Interestingly, my own research gradually moved in a different direction. Whether today's systems ultimately deserve to be called "intelligence" is certainly an important debate, but I found myself increasingly asking a different question.
How should democratic societies prepare for profound technological and social change—regardless of whether that change is driven by AI, robotics, biotechnology, demographic shifts or other disruptive forces?
In that sense, artificial intelligence became less the destination of my research and more the starting point. It exposed broader questions about institutional adaptation, democratic resilience and society's capacity to manage accelerating transformation.
I hope you'll continue to challenge these ideas in the next series. Thoughtful comments like yours help ensure that the conversation remains grounded in both technical reality and broader societal implications.
Well, the likes of Elon Musk, Peter Thiel have a very different idea about the use of AI. Transhumanism. Techno-feudalism. The opposite of Democracy. Elite theory says The People will always lose. It is a matter of which elite will rule.
And outside of direct Democracy, greek style, the spread of democracy has been very good for ruling elites. Less need for direct control. It is what globalism finally achieved. The positive capitalist story. But it has a shadow side, just like Socialism.
Thank you for raising an important point. Ultimately, every technological revolution is also a question about power—who benefits, who bears the costs, and who shapes the rules under which new technologies develop.
My own interest has gradually shifted in that direction. Less towards the technology itself, and more towards the capacity of democratic societies to govern periods of profound transformation in ways that remain accountable, legitimate and centred on the public interest.
Whether one is concerned about concentrated corporate power, state power, or emerging technological elites, the underlying challenge remains remarkably similar: how do democratic institutions retain the capacity to act effectively while preserving liberty and public trust?
Those are questions I hope to continue exploring in The Human Transition Papers.
Thank you, Leon.
Just to add some (respectful!) points about "AI" from my own perspective (including degree-level studies in computer science, philosophy of language, and linguistics). Computationally, at core it can be reduced to pattern-matching models with language models bolted on. The former models have been around almost since computers themselves, and the latter since Noam Chomsky's transformational syntax algorithms (which I was computerizing / commercializing in the 1980s). Since then, it's become what I liken to the technique used by both the old TV show "want to be a millionaire", and how IBM built a machine last decade to beat the Go world champion in Korea: "Ask the Audience". No shame in that, so long as you don't have the hubris to call it "intelligence".
As such, it's necessarily backwards-looking, manipulable (and extremely manipulated - which hilariously can be done even by individual users), and permanently regressive to the lowest common denominator of past human communications. As to individual effects, the more it's used by humans, the more it destroys innovation and indeed intelligence. I avoid *deliberately* using it (even for coding) because repeated use will simply make one's thinking processes fat, lazy, and stupid. There remains much information (even including old and uncontroversial literature, biographies, etc) that for some reason appears never to have been fed into it as "training data", whose effect is further to mislead users. In other words, it *can* be useful in much the same way Wikipedia is useful, though Wikipedia's manipulations are more obvious.
On the plus side, it's *not* subject to so-called "hallucinations": these are simply the algorithms working, *as designed*, to be "helpful". Also, for myself I have found it useful as "Automated Indexation" i.e. to find sources more easily. I trip on such indexation by default when typing stuff into "search" engines (strictly speaking, traversal of pre-sorted data), which is good for research. But even there, it's limited only to sources that have been fed into it by way of so-called "training data" - (see my comment earlier).
As for impact on human society, economics, etc, which I take to be your focus here, I have no crystal ball. Even after the current Ponzi-scheme-to-end-all-Ponzi-schemes collapses, It'll still be around - though hopefully affordable for (and thus accessible to) only state actors and larger corporations - perhaps even for China's offerings. That will be helpful. I do suspect fields such as education will take a long time to recover. And I do hope it continues to smash intellectual property, which has become nothing more than a rentier economy.
HTH
Thank you, Robert, for such a thoughtful contribution. One of the things I have appreciated throughout this series is hearing perspectives from readers with deep technical expertise.
Your observations about pattern matching, language models and the limitations of current AI systems are an important reminder that public discussion often blurs the distinction between technological capability and technological hype.
Interestingly, my own research gradually moved in a different direction. Whether today's systems ultimately deserve to be called "intelligence" is certainly an important debate, but I found myself increasingly asking a different question.
How should democratic societies prepare for profound technological and social change—regardless of whether that change is driven by AI, robotics, biotechnology, demographic shifts or other disruptive forces?
In that sense, artificial intelligence became less the destination of my research and more the starting point. It exposed broader questions about institutional adaptation, democratic resilience and society's capacity to manage accelerating transformation.
I hope you'll continue to challenge these ideas in the next series. Thoughtful comments like yours help ensure that the conversation remains grounded in both technical reality and broader societal implications.
Leon
Well, the likes of Elon Musk, Peter Thiel have a very different idea about the use of AI. Transhumanism. Techno-feudalism. The opposite of Democracy. Elite theory says The People will always lose. It is a matter of which elite will rule.
And outside of direct Democracy, greek style, the spread of democracy has been very good for ruling elites. Less need for direct control. It is what globalism finally achieved. The positive capitalist story. But it has a shadow side, just like Socialism.
Thank you for raising an important point. Ultimately, every technological revolution is also a question about power—who benefits, who bears the costs, and who shapes the rules under which new technologies develop.
My own interest has gradually shifted in that direction. Less towards the technology itself, and more towards the capacity of democratic societies to govern periods of profound transformation in ways that remain accountable, legitimate and centred on the public interest.
Whether one is concerned about concentrated corporate power, state power, or emerging technological elites, the underlying challenge remains remarkably similar: how do democratic institutions retain the capacity to act effectively while preserving liberty and public trust?
Those are questions I hope to continue exploring in The Human Transition Papers.