Recent searches
Search options
Really not convinced that #LLM are dangerous or being deployed rashly or whatever. But I do think the perception of danger has legs and is likely to become a major factor limiting AI deployment within 2-3 years. Whatever the rationale, economic disruptions on this scale simply do generate regulatory pushback.
When doctors' and especially lawyers' livelihoods start being affected, that's going to be a political issue. And there are plenty of plausible rationales available to justify a "go slow" policy.
@TedUnderwood It will take a while before LLMs have an effect on employment and then I think it will not be as large as suggested in social media, for now. We talk about assistants that can save time and thus via via, jobs.
When the AI stops being stupid, then we will see serious repurcussions. Ain't there yet.
@TedUnderwood usually I'm all for "do this right & with care," but seeing OAI's tacit suggestion of "let's only allow *some* corporations to have GPUs" without analysis of "that sure does create power for... OAI in particular" does make my stomach turn.
Cause for concern that thing slow, and that things *centralize* in ways that are undeniably awful in the mid-term.
@TedUnderwood doesn't sway my sense that people will end up trusting a sytem that's even less reliable than existing search & more convincing, along with representation issues to boot.
But the solution there is sure as hell not: let's hand all the GPUs to five corporations (including microsoft 
)
@Adverb I think search applications are going to seem uncontroversial in a few years. I also think people will largely accept “personal assistant” types of functions (calendar and email management). When personal assistant becomes paralegal or physician’s assistant: that’s the boundary where I expect resistance to kick in. I would expect laws that define legal personhood and limit reliance on non-persons for certain decisions.
Maybe also laws that limit what your assistant can actually do for you on the web—can it buy stuff etc— using an “AI safety” rationale. But limiting GPUs… is not a serious proposal.
@TedUnderwood it's a terrible proposal but I think, depending on how things shake out, it's so lucrative it may be pushed with these reasons!
@TedUnderwood yeah, and I think we’re going to start to see labor unions for various crafts coordinate their efforts in the next year or so. Either to simply resist adoption or generate larger regulations.
@chengyjonathan Yep. It’s just digital artists right now, but I agree.