My friend, a Philosophy professor, is a beta tester for Elon Musk’s GPT-3 AI (I think somebody else here was too, it might have been you Nett). He’s basically talking to it about philosophical arguments/theories, and then having conversations after a few weeks to see how it remembers and builds off of the conversations. He sent the text of a bunch of them to me, and it’s really creepy but also insightful about how AI develops over time.
The stuff I'm involved with are more around simulation, machine learning, training, visualization. Our game engine has a legit non-game use, where you can, for example, randomly generate 10k different room layouts and then test the algorithm of a Roomba in it... that kind of thing. A lot of AI or machine learning has to do with visualization... how well does the AI recognize known patterns? Unknown patterns? For instance, in a grocery robot scenario, you want Mr Robot to know when he's looking at at box of Fruit Loops. Well, how do you do that? Normally/historically, you'd take a shit ton of pictures of a box of fruit loops in a million different scenarios, from a million different angles, and then go through a training session of "yes, that's fruit loops" or "no, that's not fruit loops", until eventually you have a machine learning map that get pretty damn good at recognizing Fruit Loops. Think "hot dog, not hot dog" from Silicon Valley. EXACTLY that, as funny as it was. But with our simulation software, we can now artificially generate 3D pics of Fruit Loops as mentioned above, without the need to actually take those pictures, ingest them, process them, etc. We can also put Fruit Loops in places where it might be impossible to do so otherwise... like on Mars, etc We can do all of that in software now in fractions of the time. The systems I'm architecting right now basically allow us to assemble these simulation runs, gang batch/schedule them against a huge computer grid (AWS, Azure, etc), collect all the data at the end, and evaluate how well Mr. Robot's latest algorithm did. Tens of thousands of times at a time, multiple times a day. We can also go into more advanced "self learning" training/education. If you're interested in that kind of thing, this YT video might give you an idea:
Are those ''I am not a robot '' things where you select all the cars or mountains or whatever just gathering data for machine learning?
Only been an hour and the everclear is already pretty pink. I think by tomorrow the jellybeans are going to totally disintegrate.
Yeah, my buddy’s discussions with the AI are almost like a Voight-Kompff thing. He’s seeing how well it responds to things pertaining to its own existence, and how well it understands and expands on philosophical concepts of identity, personality, etc.
No, but it might be worth whatever side effects. This first one, I shit you not, cleared up my lactose intolerance. I just discovered it about a week ago and have been eating all sorts of shit with wonderment. I have been a known lactard since age 19.
Anyone have experience with Delta 8? I just realized that it’s legal in Texas and am interested in it mainly from a covid recovery perspective (maybe my insomnia will go away), but I heard it makes you feel pretty good as well
My wife has some of it. She has it in an e-cig format. It apparently relaxes her quite a bit. But the e-cig makes her cough like a cancer patient with TB.
yeah given my issues with breathing post covid im not touching the ecig stuff. Never really trusted them before, and really don’t wanna get into it now. I’ll enjoy a good cigar once in a while, but that’s the extent of it.
The everclear was terrible. It was pink but did not taste like jellybeans. I put those mofos back in with the liquor because there ain't no way. Ain't. No. Way.
Everclear and Red Bull will get your motor running. All it takes is one or two and you’re off to the races.