ACCount36

8 ms·

61.
by ACCount36 1y ago
[flagged]
62.
by ACCount36 1y ago
"If?" This thing has a goddamn LLM at its core. That's true for most advanced robotics projects those days. Every time you see an advanced robot designed to perform complex real world tasks, you bet your ass there's an L
63.
by ACCount36 1y ago
And I am tired of "mentioning the ethical and social issues". If the best you can do is bring up this garbage, then you have nothing of value to say.
64.
by ACCount36 1y ago
Your options for evaluating AI performance are: benchmarks or vibes. Benchmarks are a really good option to have.
65.
by ACCount36 1y ago
No wonder. Bluesky is where insane Twitter people go when they get too insane for Twitter.
66.
by ACCount36 1y ago
People tried VLMs on "closed set" GeoGuessr-type tasks - i.e. non-Street View photos in similar style, not published anywhere. They still kicked ass. It seems like those AIs just have an awful lot of location familiarity. They&#x2
67.
by ACCount36 1y ago
No. That's an often-repeated bullshit excuse. Payment processors have ways of passing some of the chargeback risks onto the stores, and it's not like Steam itself is chargeback central. If you just want free games, pirating them i
68.
by ACCount36 1y ago
The same applies to learning languages. You can learn a lot from textbooks... or you can use textbooks to give you the absolute bare minimum, and then just use the language itself repeatedly.
69.
by ACCount36 1y ago
The "DIY Surgical Robot" vid is so good. And it's also posted over a year ago with no follow-up. Damn.
70.
by ACCount36 1y ago
Not really a suggestion, but OpenAI has dropped some major hints that they're working on "AIs collaborating with more AIs" systems. That might have been what they tested at IMO.
71.
by ACCount36 1y ago
If it was just "noisy", you could compensate with scale. It's worse than that. "Human preference" is incredibly fucking entangled, and we have no way to disentangle it and get rid of all the unwanted confounders. A
72.
by ACCount36 1y ago
It's not too obscure. It's also about the point where some coding LLMs get weak. Zig changes a lot. So LLMs reference outdated data, or no data at all, and resort to making a lot of 50% confidence guesses.
73.
by ACCount36 1y ago
The press focus is a mix of the usual "new thing BAD", and the much more insidious PR work by fossil fuel megacorps. Fossil fuel companies are damn good at PR, and they know well that they simply can't make themselves look go
74.
by ACCount36 1y ago
In this study, it required a substantial similarity between the two models. I don't think it's easy to get that level of similarity between two humans. Twins? A married couple that made its relationship their entire personality an
75.
by ACCount36 1y ago
1. You train a model to exhibit a certain behavior 2. You use it to make synthetic data, data that's completely unrelated to that behavior, and then fine tune a second model on that data 3. The second model begins to exhibit the same b
76.
by ACCount36 1y ago
What kind of tin foil hat conspiracy land did I stumble into?
77.
by ACCount36 1y ago
It's the way it should be.
78.
by ACCount36 1y ago
In practice, we'll just let that AI have a direct internet connection, and also give it enough access to push code straight to prod. For the good measure.
79.
by ACCount36 1y ago
The last major intelligence coup CIA had (that we know of) was when the agency called the Russia's invasion of Ukraine months in advance. Going public with that was a bold call - CIA put its reputation on the line. But Ukraine was more
80.
by ACCount36 1y ago
Task length is increasing over time - and many AI labs are working on pushing it out further. Which necessitates better attention, better context management skills, better decomposition and compartmentalization and more.
81.
by ACCount36 1y ago
What? An FSD Tesla has its very own "world model". It doesn't try to reconstruct a world "photons in", from scratch, 60 times per second. It continuously updates and refines the data it already has based on the sens
82.
by ACCount36 1y ago
Your "cameras" have about 20 MP of active resolution between the two of them. With dead zones and pixels spread out unevenly. A modern smartphone has you beat. There's a small, sharp, high resolution color-enabled area in eac
83.
by ACCount36 1y ago
Statistically, SOTA self-driving cars are already superhuman. That holds for Waymo and Tesla both. They crash less, like for like, and the incidents they get into are less severe. But that's not because self-driving cars outperform a &
84.
by ACCount36 1y ago
No amount of LIDAR wankery can solve self-driving. Take any self-driving car crash where the self-driving car was found at fault. Dump the blackbox, extract the raw sensor data. What will you see? You'll see that the car had all the se
85.
by ACCount36 1y ago
Bullshit. And I am tired of having to call people out on it. Autopilot shuts down when it can't handle the situation it's in. This doesn't help it "avoid blame" at all. Because Tesla considers Autopilot implicated i
86.
by ACCount36 1y ago
You clearly have a pet issue. Why do you think that it's in any way relevant to the conversation at hand though? Do you seriously think that the main challenge Tesla is going to face when trying to scale Robotaxi up is that there isn&#
87.
by ACCount36 1y ago
That's a very safe assumption. There are more smartphones on Earth than there are humans.
88.
by ACCount36 1y ago
One use of AI tech is that it can enable megacorps to take and process actual fucking feedback, for once.
89.
by ACCount36 1y ago
Currently, there are billions of devices that are capable of storing and running a 4B LLM locally. Hundreds of millions for 32B LLMs. It would take an awful lot of effort to destroy all of that. If you're doomsday prepping, there'
90.
by ACCount36 1y ago
I'm incredibly skeptical of any non-invasive BCIs. EEG was around forever, and has completely failed to become useful as an interface system. Neuralink N1 is a fully invasive BCI, with over 1000 recording channels that go down to neuro

More ›