3 videos to think about AI
It's crazy out there. Every week there's a new AI tool promising you'll be able to automate your full life. OpenAI and Anthropic are releasing models on the same day [1, 2]. Your friendly financial advisor neighbour is convinced he needs to vibe code his own financial tracking app. How do you keep track of all that is happening? No, really, do you know how? If you do please shoot me an email! In the meanwhile, here's a few videos that helped me to make sense of what's happening right now
- Andrej Karpathy - Deep Dive into LLMs like ChatGPT
Andrej Karpathy is an AI "heavyweight"—founding member of OpenAI, former Tesla AI Director, and founder of Eureka Labs. Beside having a deep understanding on the architecture and technical details of Large Language Models and the Transformer architecture, he has an exceptional talent in explaining those concepts to a general audience. This video is a great example of it.
- ChatGPT Study Mode - Explained By A Learning Expert
Let's face it, the risk of shutting off your brain and let AI work for you is real. This is a known problem that has even a name, Cognitive Offloading. OpenAI is well aware of this, and the negative consequences it can have in the learning process. That's why they have introduced study mode, a way to use AI as a tutor rather than an all-knowing servant. Doctor Justin Sung goes in depth on whether this actually work, and how you can use this mode to maximise your learning.
- Richard Sutton – Father of RL thinks LLMs are a dead end
Richard Sutton is one of the founding fathers of reinforcement learning. In this video he shares his perspective and discuss with Dwarkesh Patel. I found it a very interesting debate, and one that goes at the core of what people think intelligence really is. If there's something I got away from this, is that LLM are very different beasts to human, and our way of learning is still very unique and not yet replicated.