Well yes but it will make search slightly easier to use for people who don’t know how to use search–while completely ruining search results, murdering the concept of truth(positive), and being a sexy chatbot boyfriend.
So who can really say whether paperclip-machining ourselves and murdering the concept of truth for imaginary bullshit that we can prove mathematically will never work is good or not?
Great talk, and very engaging on what can be a very dry subject. So many companies rushing to build AI into their products have utterly failed to consider much of this, which is one of the key factors contributing to the AI bubble.
Yeah, one of the most interesting parts for me was when he explained that summarizing using AI actually is most likely to ignore the most important parts if they’re not well represented in the training data.
GenAI is literally incapable of learning but people who don’t know how AI works believe the exact opposite. It’s magical thinking.
gaol
Edit: I learned a new word.
Yeah, I’m Australian and it looks weird to me too.
TIL
