"Mathematical genius" Terence Tao: GPT-4 cannot solve an unsolved mathematical problem, but it is helpful for work

The popular fried chicken ChatGPT has also become a research tool for the mathematical genius Terence Tao.

#Recently, he said online that he had discovered some small use cases of ChatGPT.

First, it's good at parsing code-formatted documentation (in this case the #arXiv search API) and then returning a properly formatted code query (later it Some working python code is also provided to call this API in the way I asked for, although I had to manually install a package to get it running). Second, I asked it to come up with some questions that a smart student might ask in an undergraduate linear algebra class (for which I provided some sample questions), which gave me some good examples of possible directions for the course. , as well as potential homework problems.

In summary, I have found that AI tools such as ChatGPT, while not helping me conquer an unsolved math problem, are useful for other work-related problems, although they often require some manual adjustments afterwards. .

There is not much difference between GPT-4 and GPT-3.5

How GPT-4 is good at parsing code formats document?

Tao Zhexuan asked GPT-4 to search for all arXiv articles whose metadata is a given keyword (for example, "Collatz" ) and was submitted within a certain date range (for example, between 2011-2016). According to the documentation below, what is the format of such a query?

In this regard, GPT-4 not only gives the code, but also lists the classification of the query.

A netizen asked, as a reading tool, given a paper, let it propose reviewers or speakers What questions might be asked? How does GPT-4 do in this regard?

Tao Zhexuan took it for evaluation and praised it as "really good".

"Here are the first few pages of a recent paper. Can you suggest some audience members listening to this paper? Is it a question that might be asked during a lecture on this paper?"

ChatGPT answered "Of course, there are some mathematicians here who might ask after listening to a lecture on this paper question". Then, Balabala listed 7 questions.

## Careful netizens discovered that the screenshot posted by Terence Tao was GPT-4.

then asked, "What do you think is the difference between the previous ChatGPT and the new assistant based on the GPT-4 model?"

Tao Zhexuan said, "So far, GPT-4's responses have been slightly more useful and met my requirements, but I have not seen a major difference in quality."

ChatGPT joins the workflow

With the birth of ChatGPT, many people are exclaiming that their jobs will be replaced by AI.

# Even the genius mathematician Terence Tao has started using ChatGPT.

On March 5, Tao Zhexuan said online that he decided to try to incorporate AI tools into my workflow in different ways. These include tools such as GPT-4 and DeepL.

During this period, he discovered all the hidden functions of ChatGPT, such as finding formulas and parsing documents in code format. , rewrite thesis sentences, etc.

For example, ChatGPT can sometimes do a semi-finished semantic retrieval in mathematics, that is, use it to generate some hints.

For example, Terence Tao asked ChatGPT to identify Kummer's theorem from the description. It failed to give the correct answer, but based on the approximate answer it gave (Legendre formula).

In this regard, Tao Zhexuan said that the role of artificial intelligence in mathematics is to provide a preliminary approximate answer, and then the correct answer can be easily found with traditional search engines.

##Tao Zhexuan also discovered the highlights of ChatGPT in dealing with mathematical problems. Identify transliterated versions of mathematical concepts in different languages.

Alternatively, ChatGPT can be asked to convert a bunch of references obtained from MathSciNet and put them in a LaTeX bibliographic environment Formatted as bibitems.

But doesn’t ChatGPT make mistakes?

In a question about proving whether there are infinitely many prime numbers, Terence Tao discovered that the answer given by ChatGPT was not entirely correct.

Looking at it from the other side, he found that the argumentation ideas given by ChatGPT could be fixed, and he had never seen this idea before.

##Copy, paste and complete with one click!

Netizens have also spent a lot of effort to explore the use cases of GPT-4.

#Let GPT-4 review papers not only be reasonable, but also hit the point.

"As an academic, I don't know what to think about this: I put an old paper of mine into GPT-4 (in two parts) and asked an economic sociologist to But a fair peer review.”

It creates a perfectly reasonable peer review that hits many of the points raised by my reviewers.

##For many people, if they want to do something, just ask GPT-4 directly, "Copy, paste, and... Key done!"

For example, use GPT-4 to make a complex game.

Netizens organized GPT-4, Replit, MidJourney, and Claude into an artificial intelligence assistant team and created a 3D space parkour game from scratch.

# However, I know nothing about Javascript or 3D game programming.

The above is the detailed content of "Mathematical genius" Terence Tao: GPT-4 cannot solve an unsolved mathematical problem, but it is helpful for work. For more information, please follow other related articles on the PHP Chinese website!