Mimee // smart casual dark and academic

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

The best thing I did for my intellectual development as a thinker was to keep going into rooms with people who had thought deeply about the things I care about and summon the courage to disagree with them. It forces you to become correct very quickly.

My theory about why so many on the left remain in denial about AI is that their worldview rests on a load-bearing notion of “the tech industry” as being composed of vapid morons whose accomplishments will always be superficial, never “real,” always based on some grand theft. With social media and search, the theft was manipulation of people’s minds. With Amazon it was worker exploitation. With Apple, it was a mix of these. In the left retelling of the story, no value whatsoever was created from these technologies. All a trick. With AI the “grand theft” in the telling of the left is the use of copyright-protected data in pre-training. This one is a particularly dangerous mindworm for them, since they identify with the “artists and writers” from whom they imagine this training data was “stolen.” This is why things like “mode collapse” from synthetic data, stochastic parrotry, “it can only mimic things it has seen on the web” and similar are so core to the argument for the left: it supports the notion of “tech bro” thieves—who lest we forget, and they never will let us, have no “liberal arts” training!—continuing their unbroken string of robberies. Of course the “grand theft” notion is an old motif on the left, relating as it does to a zero-sum mindset about economics, business, and growth that is. more traditionally associated with the left, though the lines have always been blurry, since the zero-sum mindset is above all else a *human* fallacy and thus a useful tactic in mass politics of all valences. The lines have become especially blurry lately, as has been widely observed. Anyway, the notion that AI *is* a genuinely world-changing technology, that it can “go beyond” its “stolen” training data, breaks this load-bearing conception of the tech industry as vapid and superficial and, more importantly, of the people within it as blood-sucking thieves.

We raised $7.5M to kill AI slop. Introducing Moda: the world's first design agent with taste. RT+ comment “Moda” and we’ll design your brand for FREE.