Stability AI CEO utters bold words: within 5 years, human programmers will cease to exist

Original source: Xinzhiyuan

Image source: Generated by Unbounded AI‌

Recently, Emad Mostaque, the founder and CEO of Stability AI, once again made an astonishing statement——

"In five years, human programmers will disappear completely."

He said that he can foresee the future picture shaped by AI. For human programmers, the prospect is really not optimistic.

**AI code generation? **

In fact, Mostaque is not just talking nonsense.

According to data on GitHub, 41% of all code is currently generated by AI.

For example, GitHub's Copilot is a big killer for AI programming.

Look at this effect, it is really one-click generation.

Even, AI has exploded in the cryptocurrency circle, and its popularity has risen.

Mostaque also said that Stability AI is becoming more and more popular. Within three months, the popularity on GitHub has surpassed Bitcoin and Ethereum, which proves the above phenomenon.

And the territory of Stability AI is by no means limited to this.

Whether it is in the field of biology, DNA analysis, or large language models, audio-visual data processing, etc., they have been involved.

When looking forward to the future, Mostaque said that the way people communicate and obtain information will undergo breakthrough changes in the future.

Speaking of ChatGPT, Mostaque believes that everyone will have one on their mobile phone next year, without the need to connect to the Internet.

Now that these disruptive AI big models can be used everywhere in the future, the way of dialogue and interaction will be fundamentally changed.

The next step in the development of Stability AI is the construction of personalized models.

This is no small challenge. The purpose is to allow everyone, every company, and every culture to have an AI system that is unique to them.

Mostaque explained that this is to achieve a key strategy of Stability AI - decentralization. This is much like the ethos of cryptocurrencies, ensuring that AI technology is beneficial for everyone to use.

At the same time, Mostaque believes that the future of AI development must be open and transparent, and technology must be fully democratized.

As we can see, Stability AI open-sourced their LLM, which has been downloaded more than 25 million times.

Stability AI's model starts with 3B and 7B parameters, and there will be versions from 15B to 65B later. And they also released the RLHF fine-tuned model for research.

Performance thieves can fight

In fact, everyone is no stranger to AI generated code explosion.

At that time, GPT-4 was released, which completely put the AI generated code in the public eye.

The new generation of Copilot X directly integrates a gorgeous chat interface, which can not only generate codes just by talking, but also talk to AI while writing codes, as well as an intelligent document system tailored for you, etc.

Specifically, Copilot X supports the following experiences: Copilot Chat; Copilot for Pull Request; Copilot for Docs; Copilot for CLI. These new features are all powered by OpenAI's GPT-4.

It's worth noting that, due to speed delays, the code completion tool is still based on GitHub's Codex model, which is trained on GPT-3.

At that time, William Zhu, a senior big data architect, said that Everything powered by AI is no longer far away.

Copilot can not only identify the code content entered by the developer and display the error message, but also conduct in-depth analysis and explanation of the purpose of the code block, and generate unit tests. You can even give debugging suggestions.

Plus, in Copilot, you don't even need a keyboard to write code anymore. Just sit in front of the computer, say "Hey, GitHub!", and start coding with your mouth. Currently, GitHub is experimenting with a new feature called GitHub Copilot Voice, a voice-based interaction system.

In addition to writing code, through Copilot Voice, you can even complete: code jump, control IDE, code summary.

Not only that, some time ago, a tweet by Matthias Plappert ignited a wide-ranging discussion in the LLMs circle.

Plappert, a well-known computer scientist, published his benchmark results on the mainstream LLM in the AI sphere on Human.

His testing is biased toward code generation.

The result was both shocking and shocking.

Unsurprisingly, GPT-4 undoubtedly dominated the list and won the first place.

Unexpectedly, OpenAI's text-davinci-003 suddenly emerged and took the second place.

According to Plappert, text-davinci-003 is a "treasure" model.

The well-known LLaMA is not good at code generation.

Of course, this is only what AI can do right now.

After all, Mostaque's rhetoric gave a time limit of five years.

Judging from the current progress, it is really not difficult to complete the AI programmer in 5 years.

References: