Happy Monday and welcome to your weekly download of AI updates.
I have little commentary to add, as this past week has been all about explainability for me. Most importantly around terms and concepts that help explain the scientific fields that together make up AI (stats, software engineering, math, etc). You can read more about these efforts at the bottom of this email, or jump right to our public vocabulary of terms here.
All of today’s links were first shared in our Artificial Insights 💬 WhatsApp Group, which you should totally join if you love the newsletter.
Until next time,
MZ
Human-AI Collaboration
Hussein Morsar, an MIT PhD candidate, explores optimizing human-AI collaboration, particularly in programming, by treating them as a single team:
My vision is that in the very near future, almost most decisions we take or even actions will be either assisted or performed by an AI agent.
MIT and OpenAI's Vision for Transformative AI
Sally Korn, from MIT, and OpenAI CEO Sam Altman engage in an insightful discussion about AI's transformative role in society.
I used to think of AI as a magical super intelligence that would solve all our problems. Now, it’s about integrating AI like any other technological revolution—hopefully the biggest and the best.
The Synergy of Neural Networks and Physical Sciences
Fascinating, technical lecture about using neural networks for simulating/validating scientific discovery. Most of it went over my head but still interesting.
These neural networks have seemed to learn something new; how can we use that to advance our own understanding?
Grafychat
Non-linear method of exploring GPT responses. Context.
Sundar Pichai's Vision for Google
I'm a fan of in-depth interviews with CEOs, and this one with Sundar Pichai is pretty insightful.
To me, we are just getting started in the beginning of what I think will be a extraordinary decade of innovation, and so I'm incredibly excited about it.
For those who like Washed Out, last week they released a pretty impressive video using Sora.
The role of education
Superb short presentation by a novelist and university lecturer in creative writing David Savill, about the implications of using AI for writing and thinking. Bonus points for building the lesson around Brazilian educational hero Paulo Freire:
From the YouTube comments:
ChatGPT is like salt. You use salt to improve the quality of food. Not eat like food :)
AI in the workspace
Here's a good overview of how AI is being used in the workspace, by Microsoft. Lots of stats and insights.
Envisioning AI Vocab
I want to share something we've been working on. When the newsletter began last year, each issue featured an emerging vocabulary, or term/concept that I had learned about and which carries a particular meaning in AI. Things like hallucination, inference & diffusion. Definitions were (obviously) created & fact-checked with LLMs.
I quickly realized that: 1) a single new term per week is insufficient to sufficiently describe everything I’m learning and 2) organizing and publishing these could be an interesting challenge. I've shared some of the previous experiments here in the newsletter, but I'm excited to say things have progressed quite a bit.
We now have nearly 450 distinct terms, generated with a CustomGPT and fact-checked with Claude. These of course vary considerably when it comes to degree of specialization.
So instead of creating a hierarchy or graph, we're thinking of dynamically selecting & ordering the terms based on your level of expertise, as explained to an LLM. In other words, the main interface will be more like a chat, and the cards themselves need to have memory of what you've read, and relationships between each other.
Here is the new public version with all relevant terms. Would love your feedback on this.
-
🏅 Forwarding this issue to colleagues and friends.
-
🎯 Commenting with your favorite talks and thinkers.
You are receiving this newsletter because you signed up on envisioning.io or Substack.