AI is trained off a massive corpus of internet exchanges. If you tell people youre going to thoroughly inspect their work or are able to check if theyre right, only people with high confidence will respond. I imagine exchanges where the OP is more involved in fact checking have a higher proportion of thoughtful and honest answers.
Much of the ‘secret sauce’ at the AI labs is not from the corpus on which they are trained but from the reinforcement training done afterwards. This step can influence the ‘personalities’ of the models and it is what makes them better at being ‘agents’, able to string together various individual steps to achieve your goal.
You could imagine that telling the LLM you have a holdout test makes it ‘feel’ more like an environment in which it was being RLed and therefore makes it better at seeking the reward by doing a good job.
That would honestly be an incredible performance art piece, like a distilled waste of a human life just to prove a point. Then even after all that you could ask the question "Is the art inferior, did it prove the point effectively.". I think there's a real argument to be made that it didn't, becuase just having the argument surfaces some very interesting points about worth.
Obviously people know, but theres no impulse to introspect on why or how. Knowing that someone else knows you had a hard braking event taps in to our social brains to provide a much stronger response to the event. When we know people are watching we're more likely to try and justify our behaviour.
In the current social climate I would absolutely not trust public media to understand general consensus. Ask specific people you trust or seek out their opinions.
In mainstream media, public consensus is bought by the highest bidder, or the whims of the board of the company.
In social media, general consensus is owned by those that control the best and most bots to direct the conversation.
Unfortunately most people are too lazy/busy to seek out trusted information, and many if not most have no ability to understand if the answer they get should be trusted or not.
> In social media, general consensus is owned by those that control the best and most bots to direct the conversation.
Isn't it owned by the owner of the social media platform? Do you think Zuckerberg, Musk, etc are neutral? There is an enormous amount of evidence otherwise.
If some bots proliferate, it's because the owners allow those bots to do so.
Listen, if they actually had the ability to detect bots perfectly just from owning a big tech company, then we wouldn't need spam filters. Perfect bot detection would be a very valuable product. It is one thing to hold responsibility to those with power, it is another to ask the literally impossible of them.
Detection doesn't have to be anywhere near perfect to be effective, though I expect that they can do it pretty well at this point. Remember they have visibility into far more than users do.
> we wouldn't need spam filters
? Spam filters rely on spam detection, and do a sufficient job.
It's important to distinguish between the Framework 13 and the Framework 16. The Framework 16 is by far the most ambitious of the two, and so it has had a lot more issues. I use a Framework 13, and I've loved it. It's light, has a solid frame, and runs Linux great. The battery life isn't great, and the speakers aren't either, but I've been able to mitigate the latter with EasyEffects.
If the 16 performs worse in the power efficiency department, that is not great, but it doesnt make my machine run any worse. Calling it heavy is crazy to me, the thing is tiny. If you think it's heavy you'd have trouble using an iPad. The screen thing was a shitty manufacturing issue, they released a kit to fix it, which I luckily didnt need since mine came after they fixed it in production.
I've got a Framework I am not too upset with, but the battery life (especially during sleep) is definitely one of my gripes. I still have yet to try powertop or other tools to optimize, maybe I would be proven wrong.
Sleep just ceased to exist in the last few years and got replaced with an always on, low power mode.
I believe the reasoning was partly that suspend to RAM had serious reliability issues due to the complexity of saving the state, partly that people starting expecting cell phone-like performance where eg, mail is always received.
None of the shadows you can physically see in the photos have 0 area because the construction of the fractal isn't perfect. What a perfect construction would look like and if it's ever theoretically possible to make physical are complicated questions. Beyond me for sure, how would an infinitely thin object even interact with light?
I would assume the problem with the idea is in the fractal physics rather than the definition of area, which has been solidly useful for me.
