3 ms·

The most popular AI model right now is LLM. Does LLM help "Social Media, Influencer culture, Creators, and a Ad-tech driven surveillance" or "Robotics, Spacesh

The most popular AI model right now is LLM. Does LLM help "Social Media, Influencer culture, Creators, and a Ad-tech driven surveillance" or "Robotics, Spaceships, and underwater cities" more?

IIAOPSW 4y ago
Robots and shit. What value is social media when its all just a big fat LLM dumping ground and everyone knows it. Long Live the Machine, killer of the bird and the book. That is my prophecy.
TeMPOraL 4y ago
That's a bit unclear at this point. Around GPT-3 it became apparent that LLMs are no longer glorified Markov chains - the training sets grew large enough, and the latent space high dimensional enough, that the model starts to capture the kind of fuzzy associations humans make. That is, it seems to capture an important subset of what we consider thinking. So I guess GPT-4 may still be most useful at destroying the modern information economy, thanks to being an infinite generator of garbage indistinguishable from content marketing. But GPT-5 might be just good enough to enable some qualitative breakthroughs in robotics, spaceships and underwater cities... (NOTE: this is not be believing that OpenAI solved consciousness. This is me recognizing the surprising ability for generative language models to navigate abstract concept space, just by virtue of being exposed to enough association implicitly encoded in everything we say or write...)
jhanschoo 4y ago
> Social Media, Influencer culture, Creators, and a Ad-tech driven surveillance I think it does help those. You've seen how the early models had to be taken off because teenagers trained models to be bigots. If there's a widely-deployed model that uses near-contemporary data, it would be the prime target for all kinds of promotion, influence, disinformation and product placement just like how Google became one such. The reasonable, useful replies that current models give you may become a thing of a past once people start to target LLMs for the same things that they today target Google's search, and you will have the same SEO arms race, albeit probably harder to combat considering how much more opaque LLMs probably are compared to Google's search algorithms. Perhaps the good useful information that we get from LLMs today are more a product of them being trained on paid content of high quality that aren't widely distributed.

The most popular AI model right now is LLM. Does LLM help "Social Media, Influencer culture, Creators, and a Ad-tech driven surveillance" or "Robotics, Spaceships, and underwater cities" more?

IIAOPSW 4y ago

Robots and shit. What value is social media when its all just a big fat LLM dumping ground and everyone knows it. Long Live the Machine, killer of the bird and the book. That is my prophecy.

TeMPOraL 4y ago

That's a bit unclear at this point. Around GPT-3 it became apparent that LLMs are no longer glorified Markov chains - the training sets grew large enough, and the latent space high dimensional enough, that the model starts to capture the kind of fuzzy associations humans make. That is, it seems to capture an important subset of what we consider thinking. So I guess GPT-4 may still be most useful at destroying the modern information economy, thanks to being an infinite generator of garbage indistinguishable from content marketing. But GPT-5 might be just good enough to enable some qualitative breakthroughs in robotics, spaceships and underwater cities... (NOTE: this is not be believing that OpenAI solved consciousness. This is me recognizing the surprising ability for generative language models to navigate abstract concept space, just by virtue of being exposed to enough association implicitly encoded in everything we say or write...)

jhanschoo 4y ago

> Social Media, Influencer culture, Creators, and a Ad-tech driven surveillance I think it does help those. You've seen how the early models had to be taken off because teenagers trained models to be bigots. If there's a widely-deployed model that uses near-contemporary data, it would be the prime target for all kinds of promotion, influence, disinformation and product placement just like how Google became one such. The reasonable, useful replies that current models give you may become a thing of a past once people start to target LLMs for the same things that they today target Google's search, and you will have the same SEO arms race, albeit probably harder to combat considering how much more opaque LLMs probably are compared to Google's search algorithms. Perhaps the good useful information that we get from LLMs today are more a product of them being trained on paid content of high quality that aren't widely distributed.