

Very little science is this exciting


Very little science is this exciting


RLHF was a fundamental mistake. Human feedback almost always trains an AI to be sycophantic because humans in general are super easy to flatter.
We are building the perfect addiction machine, far more powerful than social media is, and it actively undermines the honesty of the system.


It’s a finger under heavy magnification, if you’re not joking.
If you are, it’s because they wanted to demonstrate how e-fish-ient their new technology is.


Haha that picture shows one on a finger. It’s literally half the size of a single ridge of your fingerprint. Like a speck of dust.



The robot’s electrokinetic propulsion exploits microscale physics. Platinum electrodes drive fluid flow with no moving parts at around 60 nanoamperes. Four electrodes enable translation, rotation, and arcing. They currently operate at 1 volt, but could reach 10 times faster speeds near water’s electrolysis limit.
Electric field propulsion, apparently


It’s a marketing pitch. They’re showing why the US should buy their tech


I always assumed nanoscale robots would need to be lithography-based. Seems that was right. On the other hand, I assumed a MEMS actuator… their solid state propulsion is very interesting.


I won’t stop them, I will just socially shame them with an intent to stop them.
You have all the tact of a conversion therapy camp. I have nothing more to say to a closed-minded person like you.


No. Gatekeeping was the polite way to say that you have no fucking right to stop someone from doing what they enjoy, as long as they aren’t harming someone else. I personally don’t read YA novels, but I will protect that right for everyone else.
The correct term, rather than gatekeeper, would be fascist. You can fuck right off if you’re going to try controlling what media a person consumes. If you truly think what you’re saying, you are a morally bankrupt human.


Ew, that’s not true either. Lemmy is NOT the place for gatekeeping assholes. Take that bullshit to X.


Adults are allowed to read things written for children. Shaming people for that is ageist as fuck.
There are LOTS of good reasons to stop engaging with HP, but this ain’t one of them.
An operating system can lie about that though. The only reason it doesn’t is because of convention.
There is no technical reason it couldn’t look like a different OS. Try changing your user agent, it’s that simple in most cases.
How will they check for a compliant kernel, at a technical level? I haven’t seen any proposed way to do that that can’t be easily circumvented.
But here’s the thing, nobody knows what operating system you choose to install. This regulation will be equally as effective as anti-pirating legislation has been, which is to say, essentially nil.


Extra big ass-language model


Earth (and life) will be around long after humans have destroyed themselves.


I’m just saying that the proposed solution will be ineffective at filtering bots, but create a false sense of security.


It’s trivial to automate browser clicks like that with AI these days.
But yeah, more expensive than an API call.


If you can connect to a site using a browser, so can a bot.
Full brain and body emulation is seriously cool shit, even beyond inherently cool things like space exploration.
This is one of the first steps of how we finally learn the fundamental secrets of intelligence.