PIONEER: What is Generative AI? | David Foster

The GenAI 101 class we've all been waiting for

Happy Tuesday…!

When I started this newsletter, I promised to start with the fundamentals. I firmly believe that understanding these fundamentals will help us conceptualise the inflexion point we find ourselves at.

So I am thrilled to bring you *the* Generative AI 101. In this PIONEER interview I speak with David Foster — someone who has a knack for breaking down these complex topics in a way everyone can understand.

👨‍💻 David Foster is the Founding Partner at Applied Data Science Partners, a London based data science consultancy, and the author of Generative Deep Learning: Teaching Machines To Paint, Write, Compose, and Play published by O'Reilly Media.

🎉 GIVEAWAY: O’Reilly is offering 3 copies of David’s new book Generative Deep Learning (2nd Edition) to our readers! Support the EGAI newsletter by gifting a subscription to a friend 🎁 3 lucky gift-givers will receive a copy of their e-book once the new edition is released in May 👉 Click the link below to share a subscription:

What is Generative AI?  What makes Generative AI different from other types of AI? We discuss how advances in this field over the past decade have led to where we are today (GANs ➡️ Transformers ➡️ Diffusion models), and break each of these down one at a time. We also touch on the compute costs and model parameters of these models and speculate on their limitations. We conclude by discussing the future of AI and the impact it will have on education.

Some philosophical questions too…for example, what if we trained these models on all the knowledge in the world? 🌎 We almost are already…

These large training data sets are almost the entire collection of human knowledge

In this episode we cover:

1. What is Generative AI? [01:37]

2. Breakthroughs that enabled Generative AI [03:35]

3. Timeline on progress [05:55]

4. GANs (2014) [07:15]

5. Transformers (2017) [10:18]

6. Big model era (2020) [19:15]

7. The mass adoption era (2023–) [35:39]

8. Education [41:22]

David summarises the history of GenAI models in this neat timeline ⬇️

More PIONEER interviews incoming 🚀

Namaste,

Nina