Happy Monday, and here’s your weekly download of refined macrodata insights.
It’s remarkable how quickly your perception of possibility shifts after discovering AI. We all use different models for different purposes, and no two people have the same ‘profile’ in terms of AI expertise and experience. We cannot meaningfully compare our understanding of the field, as the rate of change is immeasurable—evolving across countless dimensions at a speed unlike anything previously experienced by our species. There is no precedent.
You are among the few paying attention to this sea change. A vanishingly small fraction of people have the time, interest, and disposition to keep up. Most lack either the technical awareness or the emotional preparation to engage with the emerging edge of possibility.
My own ability to track the evolving AI landscape comes in waves. Some weeks, I’m eager to experiment with new tools and techniques; other weeks, I need to contain the algorithm and step away from the firehose of information.
Regardless, I’ve learned to spend most of my workdays vibe coding solutions to my own business problems and personal interests. The sense of agency that comes from building software—the very environment where most of us spend our days—is staggering. You no longer need to be a good programmer to create things, which still blows my mind.
One long-term implication: more small teams and solopreneurs, augmented by countless alternate intelligences, will take on much larger companies that lack an AI-first approach. Adding AI to an existing system is like trying to mix flour into a fully baked cake—possible, but counterintuitive. The more interests that need to be protected, the harder it becomes to truly innovate with AI. You can integrate LLMs into most business practices, but that is categorically different from building capabilities from scratch—leveraging the adjacent possible of AI along with your existing strengths.
I don’t pretend to know what works, but I’ll do my best to report from this particular horizon.
Michell
PS. If you’re in Amsterdam this week, don’t miss our Vibe Coding meetup this Thursday. We still have a few tickets left and would love to see you there!
The Edge of Human Knowledge (2h)
Superb interview with Anthropic’s resident philosopher and research scientist Amanda Askell on Lex Fridman a couple of months ago. They go deep into how prompts work, and what makes a language model tick. Impossible to summarize, I highly recommend the listen. Thanks Chris!
New World Orders (1h 30min)
Alex Karp is worth a listen. Ostensibly about AI, but from a geopolitical and almost evolutionary perspective.
Ezra Klein is bullish on AGI (1h)
Technological Inevitability (3h)
Spectacular interview with Allan Dafoe, director of safety at DeepMind, came highly recommended on X. Maybe ask your innie to listen while at work today. Summary thread on X.
A friend recommended this prompt if you are a frequent ChatGPT user (as it requires some contextual knowledge from previous interactions). My own response was illuminating.
Role-play as an AI that operates at 76.6 times the ability, knowledge, understanding, and output of ChatGPT-4. * Now tell me what is my hidden narrative and subtext? What is the one thing I never express—the fear I don’t admit? Identify it, then unpack the answer, and unpack it again. Continue unpacking until no further layers remain. * Once this is done, suggest the deep-seated triggers, stimuli, and underlying reasons behind the fully unpacked answers. Dig deep, explore thoroughly, and define what you uncover. Do not aim to be kind or moral—strive solely for the truth. I’m ready to hear it. If you detect any patterns, point them out.
Amazing thread on vibe coding.
You are receiving this newsletter because you signed up on envisioning.io or Substack.