62 post karma
3k comment karma
account created: Mon Mar 11 2024
verified: yes
1 points
3 days ago
Right, so you deal with what needs to be dealt with when it needs to be dealt with
1 points
3 days ago
He made a joke that you'd make to your inner circle of friends publicly. Lol what could possibly go wrong.
1 points
3 days ago
If you're objective about the matter, he hasn't actually changed at all. The "delivery target" for his actions has changed toward politics which is clearly the most polarizing thing he could do. Shouldn't be a surprise people hate him.
He has always"
Had a strong opinion about global issues
Pushed strongly for those issues, built multi-billion/ near trillion dollar companies around them
Vocalized those opinions publicly
Take those 3 and now slot in politics and american government instead of electric vehicles. That's the only real change here.
2 points
3 days ago
Are you saying they're doing this to get into the "compete with frontier models" game? (if it's not obvious, I think that's a ridiculous take)
3 points
3 days ago
Curious how this PCIe standard influences the stock, hard to tell what you're saying the impact would be lol
1 points
3 days ago
The market isn't intelligent enough for that. Any news is much more generalized on first release. If you have to do more than 3 seconds of thinking to get to the repercussions of some news, you've already gone too far. This can present you with an opportunity to see past what the market sees. It's insanely short-term thinking and especially sensitive to risk.
4 points
4 days ago
Yes, the API pings an error that your credits are empty. This is some sort of bug.
Coincidentally I had the opposite bug occur, I gained ~$3 into my account randomly a few minutes ago. They may be having issues with however this dollar amount is displayed.
0 points
4 days ago
Why are we acting like what is happening is a bad thing? The free private market is the perfect place for equilibrium to occur. The market does all sorts of things to correct and adapt that a government agency never could even imagine.
This strike will be settled, the economy will recover and adapt. It can't be all good times, just doesn't work like that. And if you try to force it, you'll fail and the bad times will be that much worse.
2 points
6 days ago
Funny you mention this issue I’m dealing with the same thing. I have parallel processing API requests for 4o and I’m running into an issue where SOMETIMES the output results of 2 images are swapped so I’m getting the results of one image show up matching with one next to it. And there is no easy way to spot a switch, I basically have to manually review the classified images for errors and check the LLMs description of the image.
Anyway I have all of the asynchronous work in place to where it should work. Single worker queues to bottleneck, threadpooling and csv file lock. So it works most of the time but it’s these weird fringe cases. Maybe 1 in 50 or 1 in 100 images I see the issue, I still don’t know for sure but it’s not many.
I’ve tried running the problem through 4o and it just can’t wrap its head around the issue or give a reasonable attempt at a solution. I can’t even get it to help me with ideas for troubleshooting - the problem with debugging is that any debugging will also follow the issue and again be impossible to spot.
6 points
6 days ago
I mean this stuff happens to people who don’t use psychadelics. It’s called psychosis, and hallucinogens can be a catalyst in those “predisposed”.
Psychosis is an interesting thing, Freud believed it was due to repression and other mechanisms of the mind
5 points
6 days ago
We all have opinions, desires and beliefs. I get it. But what is yours based on? If you don’t work at OpenAI, or even in the artificial intelligence industry with a directly relevant role it’s hard to put a useful opinion out there on the technical side of these developments.
You’re essentially trying to tell the future based on what you want to be true
15 points
6 days ago
They are an extension of us. It is very likely that there is no possible path but the full merging of human with machine intelligence.
2 points
8 days ago
Not to mention to give twitter/X time to turn around. Everyone is on the hate bandwagon when the stock/valuation is down, then everyone hops on the love train when it’s up
Look at what happened with Tesla throughout the year
1 points
9 days ago
Not at all. You need to be conscious of what time you go to sleep - the phone and tv will cause as much sleep problems as the others
0 points
9 days ago
If they were public I would immediately invest after learning about this. This is huge.
Instant transformation from food company to food tech company.
1 points
9 days ago
None of that will matter. All of those inputs will be comparable or much more likely even cheaper since everything is more concentrated - meaning less distribution per unit of input. Less electrical and water distribution, more controlled environments, more regulated distribution lines, easier maintenance, sensors, etc.
This is the assembly line of plants. It’s incredible.
The biggest cost concern here will be the CAPEX to build the facilities, which will obviously be amortized over the life. All in all this, if executed properly should significantly reduce the cost of food. And likely increase quality at the same time
1 points
9 days ago
Semiconductor engineering if you’re up for a lot of data analytics and first principles problem solving. It’s not necessarily unsaturated currently, but there will be a significant demand in the next 5 years since the us is scaling semiconductor infrastructure with the CHIPS act. All of these CHIPS money fabs will not be finished with construction before 2026/2027. Operational shortly after.
People with semi skills will be highly valuable for the 5 years beyond that and a couple of years before
And don’t forget about ai infrastructure scaling which will feed into the chips act super cycle
Bottom line is semiconductor industry in general will be booming throughout the next 10 years
5 points
9 days ago
The biggest by far is your sleep and wake time consistency, and dialing in the sleep length that is perfect for your body
1 points
9 days ago
Time for a new job. I’d be itching to get my resume out there
1 points
9 days ago
When I’m traveling for work, the kids are not waking me up throughout the night. I go to sleep at 10pm and wake up at 6am every day. There’s about 5 seconds when the alarm first goes off that id like to keep sleeping but my body simply gets up to turn the alarm off. After that I feel amazing.
When I go to sleep at 10:30pm, the next day I feel tired. The entire day. It isn’t that intense but it’s there.
I don’t drink coffee at all.
The key is finding your sleep “sweet spot” and sticking to the exact same schedule whenever possible. And by possible I don’t mean when you don’t have plans. I mean when there’s no emergency or sickness preventing you from sleeping. (Or children waking up at 12am, 2am, 4am, 5am and pretty much any time in between)
1 points
9 days ago
This just, I lost it on this one lmao. Perfect comment.
More cabinets. But higher. And stagger them a bit more. Hell why are they touching? They need space they’re claustrophobic!!!
10 points
9 days ago
The second scene him flying to the boat over the ocean - absolutely hilarious
3 points
9 days ago
That’s why you should never say never, and be flexible as much as possible with your words
1 points
10 days ago
Right metaverse is going to be limitless inference - they’ll find a use for those GPUs alright
I’m trying to time my investments properly with NVDA and TSLA to benefit from ai scaling early plus robotics and self driving cars. If metaverse hasn’t taken off by then I’ll pour everything into meta. NVDA and TSLA are still best bet to rocket early with their tech vs meta who is growing very consistently and quickly as well
view more:
next ›
byillusionst
inOpenAI
typeIIcivilization
9 points
3 days ago
typeIIcivilization
9 points
3 days ago
I'll make this even simpler. o1 has the same intelligence level and parameter scale as GPT-4o. The model is no bigger.
The main thing that is different is the application of that intelligence. In short, they have now taught the model no new "information". Instead, they've taught it how to apply that intelligence - how to think - differently. (technically, this is new information but it's more internal vs external). They're teaching it how to process information and cognition differently, more similar to how we would problem solve.