[josters]

> ChatGPT cannot navigate the web yet, but it will soon. Combining the reading comprehension capability of modern AI systems with web navigation will allow us to have agents roaming the web and finding the information we want.

Won't these agents also be influenced or even biased by the prompts they're given? Maybe I am missing something, but when a misleading link based on a suboptimal prompt is formed and reinforced via recurring prompts, how can we make sure that we are getting back the actual connection of pieces of information?

[jasfi]

AIs automatically posting content will surely be used for SEO purposes. This is a very disheartening thing, but it's inevitable. We should have regulation to know if content was written by an AI, a human, or both. Otherwise AI will train on AI written content, what will that end with? AIs don't care if the content they train on is BS, so it would end with a very poor system.

[3dm4r]

I think we will need to create personalized agents on top of indexes. There is a principal agent problem between users and search engine robots that will become even larger as those systems become smarter.

[koolala]

Isn't it better than having humans do it instead individually. Yes, it will have bias, but it will all be biased the same from 1 models perspective. The data could then be adjusted for bias.

Humans doing it would instead create bias in many different directions. An AI attempting it is a good starting point.