PIONEER: AI for Everyone | Thomas Wolf

The power of open-source AI, Hugging Face's partnership with AWS, education, safety, and AI art

Happy Wednesday all,

I’m writing from Brussels, where I’ve just been keynoting at Cybersec Europe.

One of the key trends I discussed was the pace and scale of the GenAI revolution and its democratization… it’s not only Big Tech that’s got a role to play — we all do. The open-source community around GenAI is thriving. Guided by its mission to make AI tools accessible to all, Hugging Face, the open-source AI community it is one of the most important players of all 🤗.

To dig deeper into this democratisation of AI — I spoke with 🐺Thomas Wolf, co-founder and CSO at Hugging Face. 

We talk open-source vs closed AI, Hugging Face’s partnering with AWS, AI art, safety and risk, and how these models are playing out in the real world. 

Thomas firmly believes that AI models are best built via communities and collaborations (i.e. *open-source*) and explains his argument here for why this is the winning solution for building robust AI.  We discuss the need for better AI education, why creativity is no longer safe from automation, and the social impacts of these tools. 

Long term, open-source is always the winning solution

So many of the tools we use are built with open-source software. Thomas thinks that when you have many people working on something, it inevitably makes it better and more robust because it accounts for more ‘edge cases’. An example of this is remedying bias in AI models — when many people collaborate, unintended stereotypes in models are more likely to be noticed.

Should we optimise AI models for an idealised version of the world, or for reality as it is?

🎨 Creativity as we commonly perceive it is no longer safe from automation. Advances in AI art have shaken our confidence in our superiority as humans. Thomas phrases this feeling as: “We’re not at the centre of the universe anymore…” 🤖 Sentient AI might still be a way off, but it does feel eerily closer…

A mission to make AI accessible to everyone

In this episode we cover:

1. Hugging Face’s mission [02:51]

2. Hugging Face’s partnership with AWS [06:55]

3. Open-source vs private enterprise [09:01]

4. Safety and risk [11:27]

5. Dealing with bias, stereotypes, and real-world impacts of AI [19:09]

6. The pace of progress [23:14]

7. Thomas’ take on recent breakthroughs in AI [25:51]

8. Public perceptions of Generative AI and the need for better education [30:56]

9. Enterprise applications and real-world uses [33:20]

10. What’s next for Hugging Face [35:51]

Namaste,

Nina