3 ms·

And on top of that- more and more web content (especially dubious content) is going to start being generated by these kinds of models, which will bring down the

And on top of that- more and more web content (especially dubious content) is going to start being generated by these kinds of models, which will bring down the quality of Google results too Maybe Google starts filtering down more aggressively to only trusted sources (by domain or whatever else)- but could you do the same thing with a model like this, to improve its accuracy? Right now it's trained on the whole internet, but I doubt it has to be. At that point it really is just a competing indexing system I bet you could even train it to find and list sources for its claims

wolpoli 4y ago
We'll all have to dig out our old copy of Encarta at some point.
visarga 4y ago
> which will bring down the quality of Google results too Probably will improve quality. It reads better than the average website. They just need to enable search inside chatGPT, so it can be factual. I predict we'll start avoiding human text and preferring AI text in a few years.
- ben_w 4y ago
  "Quality" is more like a vector than a real number. Well-written, factually correct, convincing, are not necessarily the same things. Language generators like this can be specifically asked to say untrue things, they don't only get things wrong just by their lack of competence.