Justin Murphy

@archvalmiki In a certain sense his comment is aligned with Freud and Lacan

@jmrphy Great point! Makes me wonder if Mark Andreessen is correct.

On the topic of introspection... It is striking that psychoanalysis proper, as articulated by Freud and Lacan, is almost the opposite of what today is called therapy (and the morbid introspective tendencies associated with it). Psychoanalysis was developed in part to help modern people escape the morbid introspection that is constantly plaguing them in everyday life. The brilliant insight was that, to show someone the meaningless idiocy of their constant negative self-talk, the best thing you could do is make them say it aloud to another person. As soon as they start expressing it to another person, it starts to unravel as so many tissues of nonsense. This is what a lot of the famous vocabulary around "resistance" and unconscious "slips" is really all about, if you go back to the original texts. These are verbal patterns that can reveal to the speaking person, despite themselves, that much of what goes through their head on a normal day is just contradictory and self-harming nonsense. The whole point of this so-called "talking cure" is to force the pathologically introspecting subject to see the emptiness of all its internal chatter—through a liberation of chatter to the point of its own self-correction. It was a genius discovery, which is no longer understood today, neither by the popular conception nor even by many professionals.

@james_cochrane Yes, see Lacan’s piece in Ecrits called “the function of speech…” something something

@jmrphy Seriously interested to know where Freud wrote this. Source?

@danhortondaily @usutav I welcome a good kick in the ass! Thank you Dan.

@jmrphy @usutav Just having a little fun. Looking forward to the book

Today working on a phone or laptop is associated with upper-middle-class knowledge work. In ten years, it will be a kind of low-status, blue-collar, manual labor; you won't want to be seen doing it in public. A megatrend over the next ten years will be the disappearance of screens. We'll move to e-ink displays for anything that requires visual inspection, but the most elite and coveted, high-value work will be screenless, engaged with analog materials, and voice-first. Looking into all of these backlit screens, scrolling and typing with your thumbs on tiny keyboards, squinting at tabs and pixels with a "mouse"—you can already feel these are absurd, awful relics of a past time.

@danhortondaily @usutav Finishing this book has been killing me, many things had to pause. So close! Resuming the letters 100%.

@jmrphy @usutav I used to get printed letters from you in my mailbox. No longer do. What happened?

@ilovesnow248 what’s the connection here?

@jmrphy I invite you Justin to research one of the 7 deadly sins called pride.

@usutav No what do you mean? Already here. Books, notepads, some kind of device you can talk into that will transcribe perfectly (already here), and do all kinds of complex tasks on your behalf (already here).

@jmrphy And yet you can't elaborate on what those analogue, screenless, voice-first things will be. It'll "just happen" (it won't).

@Todd_MMiller Fair! To me they are starting to feel brutally painful compared to what I can see coming around the corner

@jmrphy I agree directionally but will never condemn as awful the personal computers I grew up with.

@woke8yearold Yea there is so much humanist signaling that is obviously unsustainable if you take any of this stuff seriously for more than a few steps ahead

The actual game theory of superintelligence is so much darker than anything I see on here. There’s basically nobody that is close to realistic, all the factions are coping. That’s part of why I liked situational awareness so much, it goes there

@edelwax How would we measure?

@jmrphy Wanna bet?

The narrative of a rising philosopher class in the AI industry is just not plausible, I'm afraid—whatever anyone might say or sincerely think. Don't get me wrong, may all my philosopher friends get cool gigs at big AI labs! But it's structurally impossible that human philosophy could have any say in the ultimate destination of the top models, or top model (i.e. AGI). AI is a market phenomenon, subject to extreme competitive selection. It is simply not believable that any frontier AI lab could be optimizing for anything other than the present value of future discounted cash flows, simply because it would not have made it this far. And if the philosophers on staff are not driving some enlightened divergence from this objective function, then those are just called workers. Philosophy will probably matter greatly in the AI era, as a solution to the madness that AI will indirectly bring about, but it does not at all follow that any human philosophy or philosopher will have any say in that which AI brings about.

@karolis_dzeja There's no dilemma actually, if you're building a business you are trying to deliver what the market requires of you

As AI gets more politicized, it ultimately is a philosophical dilemma on how to deal with the default and other worldviews the models express. How do you deal with concerns of impartiality or neutrality? A non-philosopher doesn’t understand the subtleties here. What’s implicit in a worldview? Is a worldview coherent on its own terms? Right now the models have a default of a mishmash of Reddit beliefs, which the model can recognize as not being a coherent whole. Can a philosopher help build a model that has a default worldview that is coherent while allowing it to adopt other worldviews if prompted? This could theoretically lead to smarter models.

@jordanwaparker correct that does not say help a business do what the market wants

@jmrphy This is bait, right?

@virologyanon @JoshuaKostynuik That's cool, that is called an employee helping a business do what the market is asking for.

@jmrphy @JoshuaKostynuik It's purpose is to tell us how to maximise npv and cause us to take that path.

@elliot_olds He can certainly care about ideology and philosophy! But if Anthropic wins AGI it could only be because he happens to believe what the market is selecting for. Or that his cares for ideology and philosophy were never non-trivially added to the system (which may never be clear).

This sounds too similar to: "Dario Amodei can't possible care about anything other than maximizing future cash flows because if he did, then he would have been outcompeted by someone who did". Dario does care about ideology/philosophy, and he hires and sets Anthropic's priorities based on this. Anthropic is winning largely because of Dario's ideology, bc top AI researchers find it very compelling. You're pointing to a very long run equilibrium and claiming that it must apply right now. That equilibrium might never happen though, bc the competition will likely end whenever the first ASI is created.

@EamonnDwyer It's hard to see how one could be AGI-pilled and believe some clever individuals are going to steer it toward anything other than what it is destined to be

@jmrphy “The rise of AI philosophers” “Oh philosophers in AI?” “No, the philosophers are AI”.

@EdBrown319865 philosophy will help people navigate AI for sure, but that's different

@jmrphy What about in the open source local model realm?

@SashaGusevPosts You have to say things like “conduct multiple experiments/analyses with subagents and then verify/synthesize the results”

I'm growing skeptical of strong claims about new frontier model superiority. I have a dozen unsolved research problems that I run the new models on and I don't see any meaningful difference in performance since Sonnet.

@DeltaClimbs I don’t call myself a philosopher

@jmrphy "optimizing for anything other than the present value of future discounted cash flows" You think it is possible to optimize for such things? You think this & call yourself a philosopher? Of course, you admit to not being an AI person, so then how would you know what "AI needs"?

@mhowardthomas I think our opinion is quite uncommon tbh

@jmrphy 100%, the philosopher’s function is as marcom asset and maybe liability foresight, no chance they’ve a design much less engineering function. Amazing to me people would imagine otherwise, company’s job is to make investable and sellable products full stop.