Information persistence hunter extraordinaire. 2x Dad, AI Guy, FP Enjoyer, Math for Fun, Tech Leadership. My views do not represent my employer, NVIDIA.
NYC
Joined September 2008
- Tweets146K
- Following3.2K
- Followers10.1K
- Likes116K
I installed a hitch on my car, as this is what ChatGPT 6.1 Sol said I should do to fix my need to carry three bikes, and then it turned out the hitch it told me to get is only rated for two bikes.
1
5
259
I'm skeptical, this inability of being able to judge reality vs fiction was one of the things they said about video games in the 90s.
playmobil did a study that found that children gen z and later have been losing the ability to use their imagination before even reaching age 10- sometimes as early as 6, which affects their ability to conceptualize fiction and make believe as separate from reality... damn...
4
227
Perhaps it is that we figured out how to loan money at very low interest rates that is some form of technology for our civilization
A faster way to get to the same understanding is to examine interest rates. For a venture to be worthwhile it should bring at least the same return as lending. Before the 13th century European rates were 20-25%; by the standards of many early modern Asian countries that was low.
1
6
402
You know the thing is for those years I didn't even bother with my 401K, but if I had that would have gone up much more than inflation in the end.
1
1
516
8yo: dad look at my brainrot meme, lol
me: you are like little baby, watch this
You know the crazy thing is, there are jobs out there where this would be a "hire right away" signal
Every year I’m shocked yet again at what ravenous little monsters squirrels turn into in the fall
1
4
565
Eventually one of the models, along with a sufficiently good harness, will be able to act as a virtual coworker. I think this is now more about out of distribution no-sense behavior like deleting the prod database, than about raw capability. Something resembling human like memory remains an issue as well.
OpenAI posted a chart for the internal model that solved Navier–Stokes (the model I call GPT-Aeon) with the NS announcement. There were no diminishing returns on test-time compute, it was still going up. I will quote from the blog:
'Since August 28 we have been training a new internal model that has exhibited unprecedented performance in our benchmarks, including mathematics. This model’s training is ongoing and its performance continues to improve.'
That was two months ago.
3
4
632