Born. Living. Improving.
Joined December 2008
- Tweets970
- Following45
- Followers8
- Likes598
When you start a project with someone else in mind, and that someone else later hurts you - then you most likely don't want to continue for a while, therefore temporarily switching to elsewhere becomes the option under consideration.
Paulius Macernis retweeted
Twenty six years to turn engineers back into people who just talk for a living.
Paulius Macernis retweeted
Replacing a one-line string comparison with a network call, a prompt, and 5M max_tokens isn't progress, it's latency and a monthly bill for something str[0] already solved.
"Can I help you not to hurt anyone", - Linkin Park (when you let yourself hear the words of the "One More Light" song in the wrong way)
If you are coming up with what you want to do with the items on the table when given a freedom to do whatever you want - you are an artist, if you can't come up with what to do and you look for instructions - you are an engineer. One invents, another doscovers.
For some reason, it's ok to ask the manager what's the next task to do while at 9-to-5, but it's out of the scope to ask people we have access to about which project to jump on next. Weird... Similar situation with planning at company vs. at home..
It looks like AI does not recognize itself. I mean,it recognize being AI,but it does not recognize it's own place in the world. It only knows as much as people in the internet told it is,but not as much as came out to the surface as self-analysis or alike. It's not in actions yet
The one who made it happen - don't know how that happened, but is ok to teach about that because "It happened to me". The one who did not make it - can't figure out why, but is feeling like in theory knows but struggles with executing, therefore has enough weight to teach.
It may be brutal but healthy, the time may come one day and you will tell to AI: "Trap me for success with X". And that may be sooo... uncomfortable, I assume. Turing test comes to my mind, do we really care about a robot in this case...
Someone: "Claude, with the resources you have dedicated for me, create the LLM model for me to download that beats the best LLM model known up to now"
Claude: OK, It's good you asked thus. Here it is: <downlaod>
Dario/Bill/etc: We need to regulate AI, it's too dangerous.
I think that is bound to happen:
- Robots will become real users, e.g. there will be open air festivals with tickets to humans, robots.
- Robots will become users of envs people were not before.
- Creating value for another person will be the next economics, as attention is now.
I wonder, what is happening if you call random doors and ask "Is there anything I may help you with for 1 EUR?". What would be the rate of legit responses be, and what would these responses be about?"... Should I try an experiment, for education/fun? :)
Paulius Macernis retweeted
Look at this thing go!
100-meter obstacle course final at the 2026 World Humanoid Robot Games
