AI skeptic Gary Marcus seems to be quite accurate in his criticisms of LLM limitations like halucinations and errors, but LLMs are still raking in obscene levels of investment. Are their any parallel approaches to AI or are LLMs the string theory of this field?
DeepMind is a fundamentally different type of AI. The Thinking Game is a documentary on DeepMind. In the UK, it's available for no extra cost as part of the Prime package.
The DeepMind approach is fundamentally more experiential. I desperately want to mention a couple of things, but spoilers!
Eric what do you think about the breach between Sam Harris and a lot of people on the right. I really enjoy his clear and rational thinking and I have to say I agree with him about Bret. I know you to also be a very rational a clear thinking person, are you willing to give your position on that whole affair? In my opinion sams point about a (very understandable) break down in trust in institutions during covid has created this totally toxic information space rife with algorithmicly boosted lies, conspiracies and distrust of experts, some of it deserved but some of it also massively detrimental to the fabric of society, something Russian, China and Iran are all too happy to foment……How do you see that issue?
I personally disagree with many of the specific impressions in this question and find Sam to be the one I mainly lost respect for, but I'd also love to see the core question "Eric what do you think about the breach between Sam Harris and a lot of people on the right."
That said, I think it'd be more suited as an interview topic where there is more time to expand on it.
Naseem Taleb mathematician - probability theorist - risk analyst - claims that Game Theory cannot be applied to predict sociological outcomes. Taleb called it the 'Ludic' fallacy. Using Bayesian (incomplete) probability analysis and evolutionary (Axelrod ) game theory "tit for tat " co-operative behaviour ( and Festinger cognitive dissonance and confirmation bias ) I claim that Taleb is dangerously wrong. And that defection ( as the dominant strategy of the self-absorbed coward) will result in a sufficient destruction of social capital and unity to facilitate the complete failure of democracy. Tell me I'm wrong and why please please please ?
It would help if you tried to illustrate how you think these dynamics are playing out.
From what I think you mean you are right that Game Theory correctly diagnoses the trap, but Taleb in right to question the predictive power of Game Theory. Game Theory doesn't always apply well to real world scenarios. The best example would be geopolitics, where if anything, Game Theory predictions have actually been worse than not using them at all, for the simple reason that groups of people cannot be relied upon to be rational as a group. The internal psychological dynamics of groups are almost always a confounding factor- and it's not a matter of individual personalities but the archetype they represent for the group. For example, following the indefatigable bellicose might seem preferable to continued group humiliation.
2013 there's a utube discussion Taleb / Kahneman. Antifragility etc. A bit of Nietsche too what doesn't kill you makes you stronger. Taleb tells you antifragile systems grow more robust under duress. I suggest there's an unseen inversion. Subversive systems grow more powerful too. Under challenge. Driven by the failure of ' existential courage'. The Orwellian lobotomised lemming is now center stage. Kahneman claims we value theory over practice - deduction over induction - and dismisses the Hume problem - suggesting the future cannot be predicted based on the past. So what ? Theory will show us where the future lies. Really ? I have an analogy I call Truth Trust and the Titanic ( or clueless and glueless ) six supposedly watertight compartments out of sixteen fill and flow over one to the other. About 35% and it's done and dusted. It's 30 degrees bow down and cannot possibley recover and refloat. Western democracies are similar and have been infiltrated by a sort of Popper derivative ( the Open Society and its Enemies 1945 ) so it follows for me in democracies where say 10% of the voters take it one way or the other predicting through Game theory the death of democracy is unnervingly easy. Sorry space is limited so it's not as fluent as I'd like but I hope this helps.
Unfortunately, the West lacks the mechanisms which rid the Chinese of the Cultural Revolution. That being said, is does look as Gen Z are wholesale rejecting the ideology. Trans/nonbinary/queer IQ has nearly halved in only a couple of years. Young adults rank inflation and economic growth over issues like social justice. They value 'grown-up' values like work-life balance over activism. There is a major return to Christianity, especially amongst the young.
The thing which might just save us is that every generation rebels against the boring cultural conformity of the generation before.
I agree that people cannot be relied upon to be rational as a group but I have come to believe they can be relied upon to be 'irrational' as a group. Dinesh said recently at Trigger he has written several books on Christianity and spent 30 years on the subject yet he finds himself expected to debate people who have seen a couple of podcasts and sneer at him and say " read the bible" moron. I find this constantly in my life. I'm surrounded by people who do not do the work yet have baseless opinions for which they demand respect. No debate. They are right and I am wrong ( de Bono)
Eric! Thanks so much for coming on this show again, and for ALL the terrific conversations you have shared with us over the years. What particular song, if any, has been looping in your mind in recent days and times? And if you are up for it can you share your thoughts and feelings on Leonard Cohen's "One of Us Cannot Be Wrong?"
It would appear that AI is beginning to have a very detrimental impact on the medical profession. Many hospitals and pharmacies are using AI to send and fill prescriptions both for medications and for proceedures. There seems to be even a growing desire to use an AI-assesement of symptoms as a potential means of diagnosis. Presumably, such an assessement would in future, supplement, if not outright replace, an actual doctor's physical exam. The problem is: this is causing a massive amount of serious mistakes to be made. I hear from people in the medical profession that many patients are being given incorrect medications as well as proceedures. This not only leads to mere annoyance on the patients' part but in some recent cases has led to severe complications and even death. Are you aware of this new phenomenon? Why do you think these things are being done? And what can be done to stop the reliance on AI destroying the health of many people in years to come? It deeply seems ironic and cruel that just as medicine has made such progress with so many diseases that it should now be rendered so inefficent in this way.
Everyone focuses on the US vs. Russia relationship to be the one to set off a bomb but which lesser known conflict/competitive pairing should we also pay attention to in terms of setting off a nuclear exchange?
Eric, you seem like you'd have significant wealth, from directing positions and a career in finance. What has been holding you back from self-financing deep dives on geometric unity by experts both you and the public would recognize as credible?
On the non-public side, what intensity of good faith challenge have you had it subjected to by people you'd consider experts? Is it unreasonable to assume that even a directionally correct hypothesis would improve significantly by such a process?
I agree with the premise in your podcast discussion with your brother on his research into mouse telomeres and the impact it being blocked has on safe drug testing. What other research are you aware of that has been blocked but has similar implications if the information was shared?
In the mid-20th century, science and art shared a common narrative: progress could destroy us, but it could also redeem us. That story gave politics and cinema moral gravity. In a multipolar, algorithmically managed world, do you see any storyteller or even political player capable of restoring that unity between moral danger and technological wonder?
Why did you never release the podcast you recorded with Michael Malice?
Eric, what do you tell your kids that they need to prepare for or think about regarding a future with AI?
-Ryan from Phoenix
i love this question. :) :) :)
I was actually kind of offended (on your behalf) that you had to go on rogan with Terrance Howard to refute his theories.
How do we get back to a place where people can better detect bullsh*t when it's broadcast on major podcasts?
I’d like to see fact checks on podcasts but it would probably take the fun out.
AI skeptic Gary Marcus seems to be quite accurate in his criticisms of LLM limitations like halucinations and errors, but LLMs are still raking in obscene levels of investment. Are their any parallel approaches to AI or are LLMs the string theory of this field?
DeepMind is a fundamentally different type of AI. The Thinking Game is a documentary on DeepMind. In the UK, it's available for no extra cost as part of the Prime package.
The DeepMind approach is fundamentally more experiential. I desperately want to mention a couple of things, but spoilers!
Eric what do you think about the breach between Sam Harris and a lot of people on the right. I really enjoy his clear and rational thinking and I have to say I agree with him about Bret. I know you to also be a very rational a clear thinking person, are you willing to give your position on that whole affair? In my opinion sams point about a (very understandable) break down in trust in institutions during covid has created this totally toxic information space rife with algorithmicly boosted lies, conspiracies and distrust of experts, some of it deserved but some of it also massively detrimental to the fabric of society, something Russian, China and Iran are all too happy to foment……How do you see that issue?
I personally disagree with many of the specific impressions in this question and find Sam to be the one I mainly lost respect for, but I'd also love to see the core question "Eric what do you think about the breach between Sam Harris and a lot of people on the right."
That said, I think it'd be more suited as an interview topic where there is more time to expand on it.
Naseem Taleb mathematician - probability theorist - risk analyst - claims that Game Theory cannot be applied to predict sociological outcomes. Taleb called it the 'Ludic' fallacy. Using Bayesian (incomplete) probability analysis and evolutionary (Axelrod ) game theory "tit for tat " co-operative behaviour ( and Festinger cognitive dissonance and confirmation bias ) I claim that Taleb is dangerously wrong. And that defection ( as the dominant strategy of the self-absorbed coward) will result in a sufficient destruction of social capital and unity to facilitate the complete failure of democracy. Tell me I'm wrong and why please please please ?
It would help if you tried to illustrate how you think these dynamics are playing out.
From what I think you mean you are right that Game Theory correctly diagnoses the trap, but Taleb in right to question the predictive power of Game Theory. Game Theory doesn't always apply well to real world scenarios. The best example would be geopolitics, where if anything, Game Theory predictions have actually been worse than not using them at all, for the simple reason that groups of people cannot be relied upon to be rational as a group. The internal psychological dynamics of groups are almost always a confounding factor- and it's not a matter of individual personalities but the archetype they represent for the group. For example, following the indefatigable bellicose might seem preferable to continued group humiliation.
2013 there's a utube discussion Taleb / Kahneman. Antifragility etc. A bit of Nietsche too what doesn't kill you makes you stronger. Taleb tells you antifragile systems grow more robust under duress. I suggest there's an unseen inversion. Subversive systems grow more powerful too. Under challenge. Driven by the failure of ' existential courage'. The Orwellian lobotomised lemming is now center stage. Kahneman claims we value theory over practice - deduction over induction - and dismisses the Hume problem - suggesting the future cannot be predicted based on the past. So what ? Theory will show us where the future lies. Really ? I have an analogy I call Truth Trust and the Titanic ( or clueless and glueless ) six supposedly watertight compartments out of sixteen fill and flow over one to the other. About 35% and it's done and dusted. It's 30 degrees bow down and cannot possibley recover and refloat. Western democracies are similar and have been infiltrated by a sort of Popper derivative ( the Open Society and its Enemies 1945 ) so it follows for me in democracies where say 10% of the voters take it one way or the other predicting through Game theory the death of democracy is unnervingly easy. Sorry space is limited so it's not as fluent as I'd like but I hope this helps.
Unfortunately, the West lacks the mechanisms which rid the Chinese of the Cultural Revolution. That being said, is does look as Gen Z are wholesale rejecting the ideology. Trans/nonbinary/queer IQ has nearly halved in only a couple of years. Young adults rank inflation and economic growth over issues like social justice. They value 'grown-up' values like work-life balance over activism. There is a major return to Christianity, especially amongst the young.
The thing which might just save us is that every generation rebels against the boring cultural conformity of the generation before.
I agree that people cannot be relied upon to be rational as a group but I have come to believe they can be relied upon to be 'irrational' as a group. Dinesh said recently at Trigger he has written several books on Christianity and spent 30 years on the subject yet he finds himself expected to debate people who have seen a couple of podcasts and sneer at him and say " read the bible" moron. I find this constantly in my life. I'm surrounded by people who do not do the work yet have baseless opinions for which they demand respect. No debate. They are right and I am wrong ( de Bono)
I had it the other day. It's a counter-productive strategy. Engaging in civil discourse expands knowledge.
You've commented a lot on your experience w Epstein. Why do you think the Epstein files are being protected by the Trump Administration?
Eric! Thanks so much for coming on this show again, and for ALL the terrific conversations you have shared with us over the years. What particular song, if any, has been looping in your mind in recent days and times? And if you are up for it can you share your thoughts and feelings on Leonard Cohen's "One of Us Cannot Be Wrong?"
If you don't "tarnish the golden rule" both of you could be right......what a wonderful world that'd be !
It would appear that AI is beginning to have a very detrimental impact on the medical profession. Many hospitals and pharmacies are using AI to send and fill prescriptions both for medications and for proceedures. There seems to be even a growing desire to use an AI-assesement of symptoms as a potential means of diagnosis. Presumably, such an assessement would in future, supplement, if not outright replace, an actual doctor's physical exam. The problem is: this is causing a massive amount of serious mistakes to be made. I hear from people in the medical profession that many patients are being given incorrect medications as well as proceedures. This not only leads to mere annoyance on the patients' part but in some recent cases has led to severe complications and even death. Are you aware of this new phenomenon? Why do you think these things are being done? And what can be done to stop the reliance on AI destroying the health of many people in years to come? It deeply seems ironic and cruel that just as medicine has made such progress with so many diseases that it should now be rendered so inefficent in this way.
How do we shrink government when it thinks bossing us around is doing us a favour?
It tells us what to think, gives us stuff we never wanted, and keeps growing like mould on a sandwich.
So, how do we convince the bureaucracy-industrial complex to eradicate itself?
Everyone focuses on the US vs. Russia relationship to be the one to set off a bomb but which lesser known conflict/competitive pairing should we also pay attention to in terms of setting off a nuclear exchange?
Eric, you seem like you'd have significant wealth, from directing positions and a career in finance. What has been holding you back from self-financing deep dives on geometric unity by experts both you and the public would recognize as credible?
On the non-public side, what intensity of good faith challenge have you had it subjected to by people you'd consider experts? Is it unreasonable to assume that even a directionally correct hypothesis would improve significantly by such a process?
I agree with the premise in your podcast discussion with your brother on his research into mouse telomeres and the impact it being blocked has on safe drug testing. What other research are you aware of that has been blocked but has similar implications if the information was shared?
Is there any indication we are coming out of our current scientific dark age? Is AI going to drag us out of it?
Hi Eric, could the IDW be brought back with an ever changing elected board as a sort of supreme court for thought?
In the mid-20th century, science and art shared a common narrative: progress could destroy us, but it could also redeem us. That story gave politics and cinema moral gravity. In a multipolar, algorithmically managed world, do you see any storyteller or even political player capable of restoring that unity between moral danger and technological wonder?