Hierarchies of Generalization
Issue 062 · July 22, 2024
Happy Monday and welcome to your regular reboot of ambitious information.
Whether we are progressing towards artificial super intelligence or not, it feels certain that the effects of AI will keep regularly advancing. In the short term we will probably feel underwhelmed and sometimes surprised – whereas in the longer term AI will significantly rewrite parts of our social order. I don’t enjoy fueling hype, yet can’t arrive at any different conclusion from observing the news reaching my filter bubble every day.
This week’s issue hopes to expand your understanding of what AI is capable of and where we are going. There are countless experts in the field, and it is sometimes difficult to navigate the ecosystem, but I hope this weekly curation of insights to be as helpful for you as it has been for me.
Until next week,
MZ
MZ
Working with AI as a writer and founder
I love deep interviews with power users of AI, to get a sense of how people who have spent significant time with these tools are developing their personal best practices for collaboration. I’m a fan of Every and thoroughly enjoyed this conversation with Dan Schipper. They mention numerous techniques worth experimenting with, including “expert prompting” and advanced ChatGPT and Claude approaches. Anyone creating repetitive texts should try Spiral.
I have a therapist, but he doesn’t know how to run a media business.
Beyond LLMs
Deep, smart, technical overview of the limitations of LLMs by JetBrains data scientist Dr. Jodie Burchell recorded at goto Amsterdam. The AI evolution timeline is especially insightful, but the whole talk is highly recommended. Superb explanation of general intelligence and how we’ll ramp up capabilities over time, including lots of practical applications of collaborating with current LLMs.
LLMs are powerful but still limited models—they're not on the brink of developing AGI but nor are they without application and only part of an empty hype cycle.
Aspen Ideas on the state of the world
Contentious is one way of describing Peter Thiel, but this interview is eye-opening. This unrelated Guardian article is a great complement to imagine how the US elections might directly affect the future of AI.
We need to find ways to do new things... if you don't allow these companies to have a certain latitude and flexibility to try to do new things, the stagnation will be worse than ever.
Runway of Power
Mesmerizing generative runway video of world leaders and their associated swag. Via Reddit.
AI heat for creatives
I should have done the same thing with my masterclass slides, but am suffering from temporary analysis paralysis.
You are receiving this newsletter because you signed up on envisioning.io or Substack.
Newsletter
Follow us for weekly foresight in your inbox.