Love that you and your friend collaborate in this way. With ChatGPT and Claude both now having memory, I imagine responses that the same question will continue to diverge for all of us. Wondering if we can prompt our way out of this?
Yes, I am grateful to have my best friend! IMHO, everyone needs a wingman!! 🙂
To your prompt question -- I like the way Peter Thiel has described it: AI will be "polytheistic". I have no doubt that despite its epistemic unsoundness, people will continue to turn to AI for one-shot answers. What LLMs spit out may be inaccurate and untrue but I think LLMs will -- for most people -- produce answers that are "true enough". I don't think prompts alone will sufficiently safeguard bc it's not in the AI companies' best interest to do that. Their interests, ultimately, is to direct traffic or beliefs in such a way that'll generate them maximal profit (eg. to the highest bidder for advertisers). So I imagine, eventually, something like that dynamic will emerge.
The major challenge for AI companies on that front, IMHO, is to do that in such a way without betraying user trust. People want the best answers; not the answers that are highest paid for. So we'll see how AI companies navigate that tension! What a time to be alive!! 🥳
Love that you and your friend collaborate in this way. With ChatGPT and Claude both now having memory, I imagine responses that the same question will continue to diverge for all of us. Wondering if we can prompt our way out of this?
Yes, I am grateful to have my best friend! IMHO, everyone needs a wingman!! 🙂
To your prompt question -- I like the way Peter Thiel has described it: AI will be "polytheistic". I have no doubt that despite its epistemic unsoundness, people will continue to turn to AI for one-shot answers. What LLMs spit out may be inaccurate and untrue but I think LLMs will -- for most people -- produce answers that are "true enough". I don't think prompts alone will sufficiently safeguard bc it's not in the AI companies' best interest to do that. Their interests, ultimately, is to direct traffic or beliefs in such a way that'll generate them maximal profit (eg. to the highest bidder for advertisers). So I imagine, eventually, something like that dynamic will emerge.
The major challenge for AI companies on that front, IMHO, is to do that in such a way without betraying user trust. People want the best answers; not the answers that are highest paid for. So we'll see how AI companies navigate that tension! What a time to be alive!! 🥳