The number of stars is nearly 80,000, and the star rating of popular AutoGPT exceeds that of PyTorch. Netizens: See its limitations clearly

As if overnight, a new top player appeared in the AI ​​circle: AutoGPT.

As the name suggests, AutoGPT is an autonomous artificial intelligence. If a task is given to it, it can autonomously propose a plan and then execute it without human intervention at all. In addition, it has Internet access, file storage using GPT-3.5 and summary generation.

For example, the user asks AutoGPT to build a website, and the request is to let it create a form, add the title "Made with autogpt" to the form, and finally change the background to blue , it takes less than 3 minutes, and without human intervention, AutoGPT does it by itself, as shown below. During this period, the React and Tailwind CSS used by AutoGPT were decided by themselves.

# Looking at an example, AutoGPT can already check information online, use third-party tools, and operate your computer. Since its launch, the project's popularity has not diminished. As of today, AutoGPT's GitHub Star volume has reached 78k, soon approaching 80k, exceeding PyTorch's 65k.

##AutoGPT address: https://github.com/ torantulino/auto-gpt

##PyTorch address: https://github.com/pytorch/pytorch

You must know that AutoGPT is a project that has just been online for a few days, and the initial version of PyTorch can be traced back to 2018. Not only that, judging from the summary of Twitter users, AutoGPT has also exceeded the number of stars of projects such as Bitcoin and Django.

## Source: https://twitter.com/MattPRD/status/1647653434760515584

Even Andrej Karpathy, the former Tesla AI director who just returned to OpenAI, commented on this: "AutoGPT is the next frontier of prompt engineering."

# However, unlike those who are optimistic about the development of AutoGPT, Jim Fan, an AI scientist from Nvidia, poured cold water on it.

Jim Fan said that he only regards AutoGPT as an interesting experiment, nothing more, and although this research is hot, it does not mean that it can be put into production. There are many cool things on the Internet The demos are carefully selected.

Subsequently, Jim Fan also said that in his experiments, "AutoGPT can solve some simple and well-defined tasks well, but most Most of the time, AutoGPT is unreliable for really useful, harder tasks.

This unreliability can be attributed to the inherent limitations of GPT-4. If you don't have access to GPT -4 weight or better fine-tuning, I don't think the problem can be fundamentally solved by just hints and tricks.

It's like there are no hints that can turn GPT-3 into GPT-4 Capabilities equal, I don't think AutoGPT's frozen GPT-4 can reliably solve important complex decisions. The current media hype is pushing the project toward completely unrealistic expectations."

A lot of people agree: AutoGPT is too limited and cannot solve any business problems

Jim Fan’s views are agreed with by many people. Some people think, "It is true that AutoGPT is a great experiment and will lead the wave of doing many cool things autonomously through intelligent agents. But it cannot become a product that can build the basis for solving any business problems. After all, it is too unpredictable."

Talk without practice is unconvincing. Someone showed up and said that he had been letting AutoGPT open a docx document and its exported files all Saturday. ChatGPT conversations to provide more context (json), explore other technical content and rewrite docx documents. Unfortunately, AutoGPT doesn’t even come close to achieving these goals, so give up.

There are many examples of this kind of experience. Some people try a lot of prompts for real-world problems, but AutoGPT always leads to meaningless differences. direction development.

Dissenters: Although it has been exaggerated, its prospects are comparable to GPT

While many people agree with Jim Fan’s point of view, Others pointed out that although AutoGPT has definitely been exaggerated and is now very "brute force" and inelegant. But the prospects it shows are still very powerful, almost on par with the GPT model.

Some people analyze the shortcomings of AutoGPT from an application perspective. Although it currently cannot solve many things well, such as loops, tangents, Complete different tasks randomly. But the thing to be clear about is that AutoGPT requires a lot of brain power, and it's expected to get better and better.

Those who hold the above point of view are not alone. "AutoGPT will definitely become more perfect as time goes by. A project like this has become a possible, although reliable use on any general purpose domain may only come in years rather than months."

Heart of the Machine Readers of , do you think AutoGPT will be a flash in the pan? Are you optimistic about its prospects? Please leave your own opinions in the comment area!

The above is the detailed content of The number of stars is nearly 80,000, and the star rating of popular AutoGPT exceeds that of PyTorch. Netizens: See its limitations clearly. For more information, please follow other related articles on the PHP Chinese website!