Happy Monday and welcome back to Artificial Insights, your weekly deep dive into the articles, interviews and presentations that will help you understand our inevitable transition towards greater forms of intelligence.
With this newsletter, I try to strike a balance between covering important releases and taking a step back to look at the big picture. I look for conversations and observations which manage to speak with both sides of the equation, avoiding to get caught up with the minutiae of technical implementations or the profound ethical effects caused by AI, which is better left to experts.
I love learning from the deeply knowledgable as much as I enjoy navigating and sharing their insight. No individual is an expert about all aspects of AI. If we consider AI as the intersection of comp sci, math, stats, linguistics, ethics, data science and psychology – then none of us can be considered true experts of the hyperobject that is AI. Irreducibly complex and unprecedented, nobody really knows what is about to happen, meaning each of us are uniquely positioned to ride the wave.
As always, it’s a privilege to have readers such as yourself with me on this journey.
Announcements:
-
Next week Thursday Sep 26 I’ll be at the Eye Museum with my friends at TrendWatching at their upcoming Trend Seminar. I’ll be presenting the latest iteration of the AI Masterclass live. Ping me if you’re around, and if you’re thinking of going, use code
ArtInsightsTW100for a special reader discount. -
Next month Thursday Oct 10 we’ll be organizing the first newsletter meetup for readers in Amsterdam. I’ll share more details in upcoming issues, but you can RSVP now.
-
Finally, on Monday Oct 21 we’ll be hosting our very first online meetup for all readers, likely designed around a recording of the masterclass. I’d also love to host a show-and-tell of AI projects from the group, so DM me if you have something to share. You can RSVP now.
As always, make sure to join our amazing WhatsApp community to be the first to know.
Until next week,
MZ
Is O1 reasoning? (90 min)
Deep and technically detailed dialogue between MSLT podcast hosts Dr. Tim Scarfe and Dr. Keith Duggar about O1’s models and capabilities. Highly recommended to understand the direction things are taking.
Teaching computers how to talk
Don’t miss Jurgen Gravestein’s deep dive into whether O1 delivers on System 2 thinking using CoT. Always worth your attention.
[
System 2 Thinking
Summary: AI labs are eager to solve System 2 thinking in AI in the hope of alleviating some of the inherent flaws of LLMs. Doing so could potentially unlock better planning and reasoning capabilities. OpenAI’s latest model release, o1, is touted as an early attempt at that…
Read more
a year ago · 3 likes · Jurgen Gravestein
ASI in 10 min
IBM is on a roll with their YouTube videos. Informative and sufficiently detailed.
You might have already watched some of the videos from this quiz, but here's an excellent test to determine if you can discern AI generated video from recorded material in NYT.
Apple Intelligence system prompts
Running on device. Funny how Apple has to resort to the same techniques as everyone else.
Do not hallucinate. Do not make up factual information.
DHH making cold hard math
Let's return to that $200,000/year programmer. You can rent 1 AMD EPYC core from Hetzner for $55/year (they sell them in bulk, $220/month for a box of 48, so 220 x 12 / 48 = 55). That means the price of one biological core is the same as the price of 3663 silicon cores. Meaning that if you manage to make the bio core 10% more efficient, you will have saved the equivalent cost of 366 silicon cores. Make the bio core a quarter more efficient, and you'll have saved nearly ONE THOUSAND silicon cores!
AI and Humanity (25 min)
Surprisingly good mini documentary about our trajectory towards AGI from Bloomberg Originals.
The Relentless Pursuit of AGI (50 min)
Insightful and fairly technical interview with Magic's founder Eric Steinberger about our probable trajectory toward AGI.
When something like AGI becomes an executable to-do list, you realize the future is closer than you think.
Snippets about O1
These enhanced reasoning capabilities may be particularly useful if you’re tackling complex problems in science, coding, math, and similar fields.
Instead of blurting out an answer right away, ChatGPT can now think through it first. The best analog is that ChatGPT is evolving from using only System-1 thinking (fast, automatic, intuitive, error-prone) to System-2 thinking (slow, deliberate, conscious, reliable). The allows it to solve things it couldn’t before.
O1 First Impressions (25 min)
If you haven't spent time with O1 and want to know what all the fuss is about, here is an excellent primer.
It’s never a simple binary. We’ll see edges of great abilities and others that are underwhelming.
You are receiving this newsletter because you signed up on envisioning.io or Substack.