was listening to Kyle Vogt about his new bot company and he described that folding laundry is sort of a frontier problem for robotics and we are still many ways out from there. There's solution from physical intelligence and probably other companies; but they are still fairly complex and not as easily reproducible
https://www.youtube.com/watch?v=eXbrt_2Fvgk
haha - yes - sometimes when you look at all the Web 2.0 and beyond startups, all of the startups seems to be at least original founded to solve problems single people think need solving:
* finding good looking people in college
* sending dm's to other people
* carpooling with strangers
* crashing on strangers couches
* getting takeout delivered
* robots/drones that fight each other
* the meta verse
* the equivalent of digital beanie babies
Meanwhile automation to help out with stuff like laundry, dishwashing, cooking home cooked meals, other household chores, helping the elderly, etc. remain untouched.
> Meanwhile automation to help out with stuff like laundry, dishwashing, cooking home cooked meals, other household chores, helping the elderly, etc. remain untouched.
What type of proof would you like? An official statement by Pakistani militias that the government asked them to do it? We see similar tactics being used in Syria by Al-Jolani / Al-Sharaa to wipe out the Alawaites (20-50,000 executed) with the same being planned for the Druze. That way the government can say "it wasn't us", although of course, they didn't prevent it.
Proofs have been provided in lot of previous terror attacks like 26/11 Mumbai attack, Pathankot, Pulwama etc. What did Pakistan do with the proofs? Nothing.
As somebody who works along with Applied Scientist helping them with tasks related to model training and deployemnt; how does one get exposure to more lower level engineering work like optimization, performance etc.
We have an ML infra team; but their goal is building tools around the platform, not necessarily getting workloads run optimially
Brendan Gregg's work on system performance and profiling is a good place to start. A lot of ML perf boils down to Linux perf or what the heck is happening in an HPC scheduling system like SLURM.
https://www.brendangregg.com/linuxperf.html
He is probably right to some extent, but he didn't solve self driving. He left his own company claiming that the problem is solved and only the boring part was left.
He also claimed that he can fix the search in Twitter, but he left after 2 months and now he is working on something else.
You might be interested in https:// www. jwz.org /gruntle/ rbarip.html (I'm obfuscating the URL a bit because the website's owner does something rude if your browser sends an HN referer header)
I believe so but from what I recall, he still wanted syncing between phone/computer which required the messages be stored at some point, alebeit with an extremely small ttl (minutes? seconds?). This miniscule retention meant the record was created despite a low ttl and so when the record was removed, it became a violation of the court-ordered record retention. Again, this is all just from fuzzy memory but I think that's how it approximately went down.
was listening to Kyle Vogt about his new bot company and he described that folding laundry is sort of a frontier problem for robotics and we are still many ways out from there. There's solution from physical intelligence and probably other companies; but they are still fairly complex and not as easily reproducible https://www.youtube.com/watch?v=eXbrt_2Fvgk
Still looking for the LLM moment in robotics