This week - Sergei Brin is back at Google to help out with AI; Google's Gemini AI might be out this year; Glorbo hoax tricks news bot; a BCI that unfolds in the ear; and more!
Hey Conrad, I'm not sure these commitments are very durable. They seem more on the order of cosmetics or optics- a form of AI virtue signalling. Even if the companies (or some fraction of them) act in good faith and make an effort at implementation, the technical problems are formidable, the company bottom line in competitiveness vs. rivals will always have the veto power and, most ominously, the current state of the art of LLMs is such that for the first time in computer science history, we have created something of such complexity that NO ONE understands what is going on ,"under the hood" of these things. So NO ONE understands what kind of emergent behavior they may display. So finally, color me skeptical.
Hey Conrad, I'm not sure these commitments are very durable. They seem more on the order of cosmetics or optics- a form of AI virtue signalling. Even if the companies (or some fraction of them) act in good faith and make an effort at implementation, the technical problems are formidable, the company bottom line in competitiveness vs. rivals will always have the veto power and, most ominously, the current state of the art of LLMs is such that for the first time in computer science history, we have created something of such complexity that NO ONE understands what is going on ,"under the hood" of these things. So NO ONE understands what kind of emergent behavior they may display. So finally, color me skeptical.