Am I supposed to be impressed? I have a PhD level intelligence and I am not exactly impressive.
post
To obtain a PhD, you need to contribute something original to your field of study, not just regurgitate what you've scraped from other studies.
GPT5 is gonna monkey a new hamlet? ITS GONNA BE THE BLURST OF TIMES?
ceos are so obsessed with this and thinking it can replace doctors in diagnosing people too.
Maybe not replace, but some flavour of AI is already pretty good at analyzing patterns on x-ray images and stuff like that which might be significant help to doctors in the future. Obviously not the glorified autocorrect Altman is running with hype-money, but actually useful neural network things (or whatever they really are, I'm not one building them).
Narrow models trained on a task specific data set tend to be very good at their specialization. So protien folding, or material sciences have benefitted from machine learning, but we shouldn't mistake that for being the same thing as chatGPT.
One of the bigger problems we have with AI at the moment (in my very inexpert opionion) is that they seem to be trying to throw LLMs at every problem and swearing that it'll achieve AGI soon.
Meanwhile Alpha Fold is more closely related to stable diffusion than it is to ChatGPT.
Also a PhD is an expert in a hyper specific niche area of their specialty. Would I trust someone who has a PhD in astrophysicist with an expertise on black holes. When it comes to talking about black holes? Yes.
Would I trust that person to give me medical advice? Probably not. Would I trust them to help me show basic car maintenance? Maybe and only because they have experience with car maintenance not because of a PhD.
I wonder whether any LLMs are any good at hypothesis generation
The word 'intelligence' doing a lot of heavy lifting in that headline.
Just a few... more... trillion dollars! And all your drinking water.
And slave labor, not only to build the hardware, but to view the mountains of CSAM and other heinous material to build the "guardrails"
Intelligence or Hallucinations?

What drug do I need to take to get PhD level hallucinations? The ones I've had so far are at best GED level.
PhD is not even a "level of intelligence".
Most people have PHD intelligence. They just don’t have the motivation, need or care to do all that fucking work to get it.
To be honest, as I chat with more and more random strangers these days it does begin to dawn on me that we all do roughly spend ~70 years on this planet devoting our attention to one thing or another, and that though people might not have what is seen as "classical intelligence" (i.e. high IQ's, political savvy, high empathy / sociopathy, etc.), we are genuinely absolute genius's in one particular field or another.
For example, I had an old roommate whose politics would make me drink and stare at the horizon whilst he consistently acted against his own self interests to punish people he was told are responsibly for his financial lot in life. But, he was an absolute wizard when it came to predicting the outcome of a sports game. It could be anything - football, hockey, tennis, whatever - he immediately ran their stats straight off the top off his head, summarized their strengths and weaknesses and came out with an outcome that was on the whole close to the truth.
Another example, my ex. We never really had deep philosophical discussions about the state of the world, and her consumerist lifestyle was one I tried to actively ignore. But, she was incredible at turning a house into a home -- her interior design skills would genuinely surprise me at how well-thought out and in-depth they were, not only in terms of style and decor, but also in the way that she would execute and coordinate the tasks with me to beautify our home.
TL;DR -- I do really think most people have high intelligence in one specific field or another, we just value people unequally using classical measures of success (wealth, education)
If your roommate was so good at predicting the outcome of sporting events, couldn’t he have used that skill to fix his financial situation, rather than blame others?
Tbh, transforming your talents into money is also a skill that some people have in spades and some people lack entirely. I've actually always bristled about that skill specifically, because I think Kim kardashian is an absolute phenom in that area and it always rankles when people say she’s stupid. She turned: a moderately famous but deceased and no longer relevant dad; relatively very high wealth (but not comparable to her current estate); an assistanceship to Paris Hilton; and a sex tape into an absolute empire. That’s a lot of points in her favor, but she makes the best possible decisions so consistently, she’s got to be one of the marketing greats.
80% prediction rate isn't anything to bank on :P
... or the money to go on that journey
Maybe it's just my field, but every PhD program I've seen, applied to, attended, sent students to, etc... was basically paid for, outright. Mostly it's a matter of moving, which is a gigantic bitch.
Paid PhDs are only the norm in stem, and those are the exact subjects where academia is a huge pay cut compared to industry. Hell, I'll be taking a huge pay cut (in terms of net hourly wage) when I finish my master's and quit my part time job, that requires a bachelor's, and start a PhD.
Depends in if you include behavior and psych as STEM, I guess, then.
Yeah, should've specified, hard sciences
You already said STEM. Some universities I've worked with include psych in the STEM department. It is a science. Fuck saying "hard sciences" like some kind of tiered distinction.
And, to my previous point... every psych PhD I've come across has been paid for. Hell, my advisor even had the balls to say "if you're paying for a PhD, you're doing it wrong," when someone asked about funding during the interview.
It's a colloquial term. My best friend is a psychologist and she taught me that distinction (I'm not a native English speaker), and I genuinely didn't know some people took offense to it. Never meant for it to be tiered. I know psychology is a science, and a natural one at that. You're the one acting like your field is somehow special and better than others. I tried to be general and you said your field doesn't fit in, so 'you already said stem' makes zero sense.
Either way, I never said it was normal to pay for a PhD? I said it's a huge pay cut vs working and industry job, which not everyone can take. Some people have others financially depend on them, and they can't just decide to accept eating half of what they could otherwise for self fulfilment purposes.
It's paid, but (at least in my case) doesn't pay that much. It's barely enough to live off of if you're really careful with your money. I wouldn't have been able to do it without accumulating significant savings beforehand.
Don't you need a Graduate degree AND a Post Graduate degree to even be eligible for most PhDs?
Me: “Hey GPT-5 Ive been diagnosed cancer.” GPT-5: “Have you thought about using cocaine and essential oils!”
Why is that image giving the same vibes as:

My brother, who has a Ph.D. in pure mathematics, says that all a doctorate means is that you've learned to talk a lot.
I don't and never will have one, so this is the only relevant information I have.
AGI isn't a myth, it's just never going to happen with an LLM as its core.
I hate the way the media makes this problem so much worse by incorrectly describing LLMs. They can't "have intelligence". They are incapable of any kind of thought. The "intelligence" of GPT1 and GPT5 are the same, in that neither have any. They are complex computational algorithms designed to generate text from prompts. That is absolutely not the same thing as thinking or knowing things.
There are entire cults springing out of the ground believing LLMs to literally be thinking feeling beings 💀 we are so beyond fucked.
You can say the same thing about an ant or a slug. I don't think the philosophy of what intelligence is is as cut and dry as what you say.
I agree they're pretty stupid, but I wouldn't say they're zero on a scale of zero to human. If an llm type algorithm happens to be some part of the human intelligence algorithm then an llm has some fraction of intelligence
PhDs themselves aren't very smart. They're just sheets of paper. 🤷♂️
I know at least one Ph. D. where this could very well be true.
Sam Altman wants to raise babies using AI.
Never forget.
Based on the fact that they'd give someone like me a PhD, this comes as no surprise. But it's not saying as much about GPT-5 as a lot of people might think.
As someone who has a PhD and self identifies as a dumbass I can confidently say, it is not a high bar
GPT5 was an effort by OpenAI to reduce costs. It is not smarter than the latest GPT4 models.
They’re also talking about data centers in space, yet are too cheap to use anything but evaporative cooling + supplemental gas generators on Earth.
I did some math on, amongst other things, launch costs for an Earth-data center sized installation, or the area needed to radiatively cool it, and it is fun:
https://projectrho.com/public_html/rocket/heatrad.php

See that power of four? Areas get very large, like kilometers wide, if you want your coolant below a typical 300K (~30C), and apparently no one told Bezos that little detail.
Those space construction startups know what they’re doing. They’re selling billionaires a bridge to nowhere; and it’s working.
They’re selling billionaires a bridge to nowhere; and it’s working.
That's just wealth redistribution /s
Those space construction startups know what they’re doing. They’re selling billionaires a bridge to nowhere; and it’s working.
Guess I should have gone CHA instead of INT.
I might have gotten some of that billionaire money to buy more RAM.
They’re selling billionaires a bridge to nowhere; and it’s working.
Look, I'm not saying it's a good thing. In fact, it would be an insanely wasteful use of resources, labor, energy, etc.
That said, folks are all about "eat the rich" and this may very well be the closest thing to that.
So for a kilowatt and 300K with a poor quality radiator (0.8)
Area = 1000/(0.8*5.670373x10^-8*30^4)
=1000/367.4401704
≈2.72m^2
So using the approximation of 1kW/m^2 of solar, you need on the order of 2.7x the area of solar for radiators
That doesn't seem too bad, and is on par with what the ISS has. The radiators on ISS have emissivity about 0.91
Ed. With the same quality of radiator on the ISS it's about 2.4*solar kilowatts
Maybe she meant PDF level intelligence
top 100 comments