Thinking Machines have put out a string of incredibly high-quality posts lately. Hard to oversell how much cred it's buying them with the AI research community! Keep up the great work folks

And this prompted me to record a video showing some of my random non-work usage recently, to give a feeling for what the app looks like :) https://youtu.be/Y2B27hdKMMA

We also showed it as part of Hamel's course: https://x.com/HamelHusain/status/1956514524628127875 (https://www.youtube.com/watch?v=DgPr3HVp0eg) which is a longer example of the tool in action

We made the tool, and that will eventually be available on its own. But the method requires some discipline and 'unlearning'. It's very hard to show someone an AI tool and not have them treat it just like ChatGPT/Claude/... - that's the part that takes the time, and having a community of people working through different examples and case studies together is a lot more motivating for this than just staring at an empty prompt box :)

I've always found this interesting. Think+transmit seems more likely to be the bottleneck vs receive, given that we can easily parse most podcasts etc at 3-4X speed. If being understood by everyone wasn't required, I wonder if one could learn to boost both send and receive rates?

This is going to be my go-to reference for 'how should I get started with electronics' - getting to see various capabilities (sense things, move motors, flash LEDs) without having to purchase parts while not knowing what you don't know is fantastic. Props for pushing through all the projects - I look forward to seeing what you make with these new skills :)

A few observations: 1) the performance vs complexity curve looks very similar to that for most humans (having seen groups attempt Towers of Hanoi with 5 car tires) haha 2) models can trivially solve some of these tasks when given tools 3) this is an internship paper with some quirks that many mostly dismissed, but is being quoted everywhere as "Apple proves LLMs can't ever reason"

Anyway, fun experiment to test your understanding of these things but don't take any conclusions as gospel :)

And Google's version: https://jules.google

Try it with your own personal photos. It is scarily good!

What percentage of human software engineers would you guess pass this bar?

I was able to read the code it wrote, and check that (as hoped) it was using a good existing library to do the heavy lifting. And I had it make plots that I could visually use to check that the values were 'reasonable'. The value in that case was simply that I didn't have to leave the couch and write the code myself (although if the result was actually needed for anything more important than a smug 'i thought so' confirmation I would still have taken over and validated it kore carefully).

I've been in similar positions. It's a good way to gain experience and level up fast. There are downsides, though - lack of technical oversight adds pressure (if I do something wrong, nobody will catch it) and lessens appreciation (nobody understands how good this is), which can contribute to burnout.

Pairing and doing code reviews together with someone, even if they're not more skilled than you, is a nice way to feel better about technical work. Catches bugs well too.

Seeing if you can learn from seniors in or outside the company is worth it. Sounds like you won't get detailed 'here's the right algorithm for X' help but that's not the worthwhile bit of any mentorship IMO - the useful parts are more often 'here's a way to push back on unrealistic requirements / work with colleague X better / handle conflict.'

Finally, remember that there aren't some mythical grownups who Do Things Right(TM) - everyone is making it up as they go along. You're already valuable to the co, and you'll keep getting better with experience - blink and you'll be the senior producing to 'highest standards' and giving all the new hires impostor syndrome ;)

Fantastic product! Just perfectly solves a real issue. Great work, I'll have to get one if any of my piano-playing family members go digital :)

I found an extra one not mentioned in the article when I tried abstaining from caffeine last week. Energy slumps mid day, a few headaches, and overall a major reduction in productivity and fun! Maybe unsurprising to most.