I hold a R&D Position at an MIT lab. I also hold gov clearances for DoD work. They are pretty accepting of the fact that a lot of folks in the field are neurodivergent. No one cares because if you deliver results you deliver results. No one cares about shit under the Trump administration because its an absolute joke that has thus far only stood to get in the way of the way we carry out research. The party of "minimal government" sure as hell loves to tell public established institutions how to carry out their own damn business.
The amount of pressure young kids are under... I am surprised the numbers aren't much higher. I grew up with debilitating OCD/Tourettes. I am glad kids growing up today have more resources than I did. Society itself is sick and broken. If that many kids are having issues.. Maybe the system is the problem here?
Ok.. Let’s ask a different question. Assuming development of super-intelligence is possible.. How do you measure it? What criteria satisfies the “this is super intelligence”? You honestly sound like most pseudo-intellectuals I hear discussing this very topic..: Ironic how you think you’re the brilliant one and it’s others who are stupid… Actually not really ironic a fool doesn’t know he is a fool.
Your employees won’t rat you out… Just don’t say “sucky” to those above you. If I have a cool ass manager who looks out for me and is real (I’m lucky enough to be at a MIT lab where everyone is cool as hell), I will always have their back…
Professional Security Researcher here.. I haven't really seen any models reliably find and exploit a 0day. Folks are are at least TRYING to develop such models internally at the MIT lab where I work, but not sure how far along they are coming yet.. If a model is developed that can find a 0day or two (like Big Sleep which I think maybe found some) I won't be surprised but keep in mind fuzzers find thousands of real 0days with far less compute... These capabilities are of course something worth looking into, but too many people are promising 0day oracles already and that simply just isn't where we are right now (or ever? ). Sorry for bad grammar typing quickly from phone here.
Again... You don't need the "best" engineers to develop some crappy derivative app that probably already exists. 99% of the stuff people are trying to build these days requires nothing more then a few competent engineers.
He wants to see a comparison and the article shows no out put comparison. It shows two inputs, says one gives more accurate output than the other without showing us...
Yea they sound like plausible enough arguments.. I often test my vague input against structured ones and for many tasks it didn't seem like hallucinations happened significantly less. Honestly the more structured you have to be, the less "general" your model probably is, and in theory you want your model to be as general as possible. Seems like here, you're simply helping it over-fit.. At least that's what my intuition tells me but I am yet to really check that out either.
Its sad content like this makes it too front page. AI has been developed and used at both companies far before LLM TRANSFORMERS made such a hit. The author clearly doesn't have even a basic understanding of what AI is... AI is a giant umbrella term. The hype is essentially all about Transformer based models and I time will show its a hype bubble they don't need to be part of. I will never understand how people with so little technical knowledge on a subject can make grandiose claims like "Apple is losing out on AI!" and be taken seriously. Every company is high on their own damn supply with this pushing AI nonsense. I don't need language models shoved in my face by yet another company. I respect Apples approach far more than everyone else trying to ride the LLM bandwagon.
Nice. I an researcher at MIT LL. In order to work at the lab you need to be a US citizen since a lot of it is federally funded. I tried to get my buddy, a Waterloo grad currently at Google, an interview but he is Canadian...
How about anyone pushing AI or AGI or anything of the sort? You think they aren’t high on their own supply? The author doesn’t hold a candle to Altman when it comes to sniffing their own farts… Cmon now. If anything it’s the opposite.
That’s exactly the problem the author has. Lying and hype have replaced genuine innovation. It’s sad that lying and pushing nonsense is “part of the game” because it shouldn’t be. The game is only of benefit to a handful of people like Altman, not the humanity or to the field in general. I am sick of it as well… Tech has become grifter central and everyone is high on their own fucking supply…
I think from the authors perspective, LLM hype has been mostly the same exact thing you’re accusing him of doing. People with very little technical background claiming AGI is near or all these CEOs pushing these nonsense narratives are getting old.. People are blindly trusting these people and offloading all their thinking to a sophisticated stochastic machine. It’s useful yes, super cool yes. Some super god like power or something that brings us to AGI? No probably not. I can’t blame him. I am sick of the hype. Grifters are coming out of the woodwork in a field with too many grifters to begin with. All these AI/LLM companies are high of their own supply and it’s getting old.
So far I’ve needed hundreds of “all you needs” beginning to think papers with that title are not all I need… Let’s try and get a bit more creative than stealing the same name scheme used when Transformers were introduced.
It's ironic because the executive aspects of the job are easily the ones almost anyone can do. I say that as someone who has created and sold startup with a couple friends. We had to take on the role of both engineers and executives. Guess which one was far far easier and which one actually mattered?
Unfortunately It seems most companies only purport to have "growth mindset" In my experience, at least now, the author nails the issue right on the head. You have executives with a dogma that ultimately means controlling others... I see little proof the opposite is true.
It's cute when people think they can all exploit the current market and "get rich" when in reality they almost always just help the same people who are already rich, get richer. I say that as someone who has himself, started and sold a startup with a couple friend... The reality is I got SUPER LUCKY. The chances of any individual "making it" in the system that currently exists are minuscule to none. You might be better off playing the lottery and hoping to hit big. The chances of either are ridiculously low. The author didn't say how to solve the bubble crisis we are watching. Usually they aren't things you can solve. He simply acknowledged a huge one is forming and he isn't wrong.
It kills the magic of coding for sure. The thing is. Now with everyone doing it, you get a ton of slop. Computing’s become saturated as hell. We don’t even need more code as it is. Before LLMs you could pretty much find what you needed on github… Now it’s even worse.
as someone who’s been coding most of his entire life I have to admit.. LLMs kind of killed the magic of programming for me… It’s not as cool when I create something with the LLM.. It just feels like you had someone else do the work for you… It’s sad kind of…
That just shows you how warped the minds of the general public are. They fail to realize the only thing Jobs did was know the right people then take all the credit…
Lmao right!? Dude stole all that from Xerox and MS came out with palm pilots early 2000s then you had black berry. Amazing how many people think CEOs actually did the work they get credit for when they almost virtually all were in the right spot with the right people. Notice none of them can reproduce or start another ultra successful startup… Zuckerberg is a perfect example of one hit wonder and he wasn’t even the first to create social sites.. You had MySpace and a ton of others..
Seriously. The transformer coupled with tons of compute is why we got here. When that paper came out and people (AI researchers) saw the results many were confused or unconvinced. No one has any clue such an architecture would yield the results it has. AI systems has always been far more art than science and we still don’t even really know why it works. I feel like that idea being stumbled upon was sort of more luck than anything…
Yea because everyone knows CEOs just take credit for the work the real engineers and others do… I’d far prefer researchers or developers getting this kind of money instead of a CEO…