Blog Archive / JACOB ROBINSON
Can I Use AI To Solve The World’s Unsolvable Problems?
[Image omitted]
Hi, it’s been awhile.
Since the release of GPT 3.5, I’ve been trying out a rather on-off little experiment. Let me show you what I mean.
Most people will agree that AI bots, as they are now, are not very intelligent. That is to say, they are somewhat knowledgeable and almost nonexistent in creativity. However, in the future (or even today, in certain combinations!) they may have the creativity to give unique solutions to already conceived human problems.
These solutions need not be correct, but rather reframe the question in ways that human actors may not have thought of, and thus lead to possible breakthroughs.
So, I’ve done two things. First, I gathered a list of currently unsolved problems. Wikipedia is great for this, as they have several listicles dedicated to that exact topic. Second, we organize all the questions into one cohesive document. Third, we go down said list and “bruteforce” answers off of AI. I would list out what LLMs I’m using here, but given how quickly the LLM landscape is changing I think its perhaps better to just have them in the core document.
Sounds fun? Want to help? Below is two buttons — one links directly to the page, where you can see what we have so far as well as engage in the comments. For those who are interested in working on the project more long term with us, the second button is to our official discord, where we have a dedicated space for talking about the project. Hope to see you around!
Select text in the article to highlight it.