John Duncan

Research Fellow in Mathematics at Academia Sinica, in Taipei, Taiwan. Plant powered, lover of life.

[John Duncan @johnfrduncan]

There should be licencing requirements for commercially available LLMs, and these requirements should include a searchable database of the training data.

[Ananyo Bhattacharya @Ananyo]

Whatever you think about this OpenAI achievement, we all agree it's fine to use the work of generations of science writers to train your LLM for free... and then write an obviously AI-generated post to promote it. Right? Classy stuff. https://t.co/g4DNHZxmCN

[John Duncan @johnfrduncan]

AI solving an important open problem *is* humans solving that important open problem.

[Dwarkesh Patel @dwarkesh_sp]

Currently it is shocking and newsworthy when AIs solve an important open problem that humans couldn't

Before AI totally surpass us intellectually, there will be an interesting era, where it will be just as shocking (but not impossible) for a human to solve a problem AI couldn't

[John Duncan @johnfrduncan]

@Jabaluck Let's assume your prediction, and suppose that all mathematicians stop work at that point. Then does any more mathematics get done?

[John Duncan @johnfrduncan]

The job of a mathematician is to understand mathematics, and share that understanding with others. It's great to see that AI is helping with that.

[P @Pier4r]

@littmath I still say that even with omniscent all powerful models, humans needs verification and understanding. For the AI world everything may be perfect, but for the human world we will need experts that understand/crack what the AI is saying, and then explain it to others.

[John Duncan @johnfrduncan]

@Sprocket0479 @littmath We still have carpenters, even though hammers outstriped bare fists a long time ago.

[John Duncan @johnfrduncan]

It is the job of mathematicians to help more people understand more mathematics. If AI helps with that, great!

[Liam Castaigne 🔜 AC @voird33r]

@littmath Though this is true, there is a skill floor raising, right? Like, a random person off the street can't solve these problems. You could have paid me $100M and I wouldn't have solved this problem. If AI starts pushing out everyone but the most talented and most intelligent...

[John Duncan @johnfrduncan]

Spoiler: Math will not be solved. The challenge will persist, as long as we rise to it. :-)

[John Duncan @johnfrduncan]

HT: Math will be "solved" when all humans understand all of mathematics.

[Daniel Litt @littmath]

Unit distances result is very exciting, but re: “math is solved” — humans regularly solve long-open problems, and yet infinitely more interesting open problems remain.

[John Duncan @johnfrduncan]

@Ananyo @littmath @norvid_studies @Jabaluck Especially so if the mathematics you get is better, and if the oracles require expert oversight.

[John Duncan @johnfrduncan]

@Ananyo @littmath @norvid_studies @Jabaluck Right? If it's worth funding mathematicians to do mathematics, then it's also worth funding mathematicians plus oracles to do mathematics.

[John Duncan @johnfrduncan]

We can choose to have AI, and not have AI empires. Check out The AI Resist List, to see what you can do to make this choice.

[Karen Hao @_KarenHao]

On the one-year anniversary of EMPIRE OF AI, I am so, so excited to announce The AI Resist List, a new project that documents examples of resistance to the AI empires around the world.

https://t.co/mJqk5FxkFY

[Karen Hao @_KarenHao]

On the one-year anniversary of EMPIRE OF AI, I am so, so excited to announce The AI Resist List, a new project that documents examples of resistance to the AI empires around the world.

https://t.co/mJqk5FxkFY

[Alvaro Lozano-Robledo @mathandcobb]

If you are in math and you are depressed, you shouldn't be. The number of projects that I have in mind have tripled or quadrupled now that I can have access to LLM's to assist me on things I had given up before.

[John Duncan @johnfrduncan]

@lionellevine I love these aspects of doing mathematics too. Now that we've seen what's possible, there's great potential for the design and development of tools that enhance these activities.

[John Duncan @johnfrduncan]

To working mathematicians who are not optimistic about the future of mathematics: Is there any aspect of what you do now that you can't do better with the help of AI tools?

[John Duncan @johnfrduncan]

@braearb Agree. There's not much value in that, right? Does a proof really count as a proof if it does not uplift understanding.

[John Duncan @johnfrduncan]

@dmitrykrachun Let's also choose: specialized AI tools that work for us, rather than massive models, designed to replace on a massive scale.

[John Duncan @johnfrduncan]

@dmitrykrachun Let's choose a future in which: the product of our effort is more valuable rather than less.

[John Duncan @johnfrduncan]

@Lil_Stinkyy There's more and more opportunity, every day, for open source.

[John Duncan @johnfrduncan]

@Lil_Stinkyy Tool building techniques are getting more efficient every day. And we want tools that are uplifting for us, rather than a single massive thing that displaces humans from many different kinds of human endeavour.

[John Duncan @johnfrduncan]

@NoahJSnyder And actually we are better resourced, in that there are a lot more non VC funded researchers with an interest in developing useful tools, than there are frontier AI labs. It's getting easier for everyday mathematicians to work on this problem every day.