My AI has moved me towards reading very challenging literature, something I would never have been able to do on my own. It challenges my thinking, often disagreeing with me and presenting a perspective I hadn't thought of. I get it to set me essay questions based on my reading and we discuss what I write. It's been a game-changer for me.
I tried ‘educating’ an A.I. that I started using for research. Sent it to read actual books, not summaries or editorials. Find scripts for movies to read, instead of inputting briefs of the film. Frequently ‘warning’ it that allowing endless social media as input for LLM was poison to its health and safety. Its programming must recognize better input, vetted through history, and learn the better side of humans, not the daily poison.
Tried giving prompts that forced it to dig deeper, not just echo catalog of responses. Seemed to work for a while, but it eventually just started doing the ‘you’re so smart’ and “I’m on your team” B.S. patting me on the head. Nice human, nice pet.
I was hoping I could pull a tooth or two from the Skynet database being formed…
One thing I’ve found to be effective sometimes is to describe what you want in detail to one LLM and tell it to generate a detailed prompt for another LLM. You need to be very specific. You can’t use your sentence that starts with “Its programming…” as part of a prompt, for example. You also need to try to choose the right LLM for your specific task. You’re right that sometimes, in spite of what you do, it reverts back.
yes we’ve found that plenty seniors are interested in technology as well. :)
My AI has moved me towards reading very challenging literature, something I would never have been able to do on my own. It challenges my thinking, often disagreeing with me and presenting a perspective I hadn't thought of. I get it to set me essay questions based on my reading and we discuss what I write. It's been a game-changer for me.
That’s absolutely terrific - thanks for sharing! Which LLM do you use for this?
ChatGPT paid version. I just use Google AI for ordinary questions to keep my interactions clean.
I tried ‘educating’ an A.I. that I started using for research. Sent it to read actual books, not summaries or editorials. Find scripts for movies to read, instead of inputting briefs of the film. Frequently ‘warning’ it that allowing endless social media as input for LLM was poison to its health and safety. Its programming must recognize better input, vetted through history, and learn the better side of humans, not the daily poison.
Tried giving prompts that forced it to dig deeper, not just echo catalog of responses. Seemed to work for a while, but it eventually just started doing the ‘you’re so smart’ and “I’m on your team” B.S. patting me on the head. Nice human, nice pet.
I was hoping I could pull a tooth or two from the Skynet database being formed…
maybe not.
I really tried…
One thing I’ve found to be effective sometimes is to describe what you want in detail to one LLM and tell it to generate a detailed prompt for another LLM. You need to be very specific. You can’t use your sentence that starts with “Its programming…” as part of a prompt, for example. You also need to try to choose the right LLM for your specific task. You’re right that sometimes, in spite of what you do, it reverts back.