Situational Awareness (058)
Your weekly exocortex download.
Happy Monday and welcome to your weekly edition of artificial insights.
Like so many others, this was an active week in the AI news cycle. In contrast, I have tried to remain focused on concentrated efforts. There is finally a first recording of the lecture which consolidates what I think about where we are heading, with a dose of practical applications for an advanced but not necessarily expert audience. Presented at a fast pace and without exercises, it runs about 90 min. I’m not even sure a “class” is the best way to convey some of these ideas, but I’ll have a lot more to show in coming weeks.
This week’s newsletter edition is packed to the brim with links to insightful articles, interviews and tech demos to help you make sense of what’s happening in the field. If you have something to share, please to so in our WhatsApp group, which is by far my favorite place on the internet to talk about and learn what is going on.
Finally, we have made stickers and other goodies for all newsletter readers who want them. If you’d like some merch in your mailbox, please share your postal address using this form.
Until next week,
MZ
AI Hype vs. Reality
There's a significant debate around whether AI is overhyped or underhyped. Current models exhibit both impressive capabilities and critical flaws, especially in general intelligence and real-world applications.
The number one concern at the moment is just AI-generated slop... it just leads to this general AI-generated miasma... a landscape where increasingly you can't trust what you see or even hear.
Subtle reminder of exactly how AI ends up biased (insight into the decision-making around tools used by billions). Google Maps decided not to implement walking direction through more scenic routes to avoid furthering social inequality.
Advancing AI for Real-World Solutions
Excellent, deep interview with Nick Frosst, co-founder of Cohere. Very technical without being overwhelming. Touches on aspects of language, AGI, training models and much more.
I think that, like, addresses this fundamental issue; now you can get a language model as an interface into some external source of truth rather than relying on the internal memory and weights of that model.
Challenging AI Limits
Detailed interview with Francois Chollet by Dwarkesh Patel about the "AGI Test" (ARC) and how comparing core knowledge with extrapolation can be an indicator of general intelligence. Great stuff.
General intelligence is the ability to approach any problem, any skill, and very quickly master it using very little data.
“Intelligence is what you use when you don’t know what to do” Jean Piaget
Anthropic just announced Claude 3.5 with Artifacts, an interface for generating documents, code, diagrams, etc. inline with chat.
Everyone aboard the hype train. Having done a fair bit of coding with ChatGPT, but never quite getting used to Github CoPilot, Claude 3.5 is genuinely exciting.
Super cool use case for co-creating software with an LLM.
Exploring AI Design with Maggie Appleton
Maggie Appleton has insightful takes on how to actually use AI in work, especially design and applications.
Condensing, synthesizing, sense-making AI is way more important and interesting than generative AI.
AI is a feature, not a product (MKBHD).
Very insightful essay about some of the long term implications of exponential new AI capabilities by NotBoring.
Tim O'Reilly on fixing copyright violations in GenAI.
This article has an unnecessarily aggressive style but some very good points about the technology hype cycle we're in.
Not strictly AI but pretty cool: Kate Crawford's new infographic is called Calculating Empires, a genealogy of tech and power since 1500.
Preview: Radar Assistant
Envisioning has been focused on integrating conversational intelligence into our research radars, and we are exploring approaches for interfacing with data visualization using AI. To showcase this, we are preparing an update to our Cities Radar with a few new functionalities, including a chat interface which allows you to navigate and rank the underlying research data with an assistant.
Reach out if you are interested in the future of cities or our solution. Lots more to share soon.
If Artificial Insights makes sense to you, please help us out by:
📧 Subscribing to the weekly newsletter on Substack.
💬 Joining our WhatsApp group.
📥 Following the weekly newsletter on LinkedIn.
🏅 Forwarding this issue to colleagues and friends.
🦄 Sharing the newsletter on your socials.
🎯 Commenting with your favorite talks and thinkers.
Artificial Insights is written by Michell Zappa, CEO and founder of Envisioning, a technology research institute.
You are receiving this newsletter because you signed up on envisioning.io or Substack.