A lot of these tools, like React, are designed to embrace, extend extinguish the web. Why Microslop and Zuckerberg spend millions of dollars of dark PR claiming anyone who doesn't like React doesn't know what's going on is because it makes the web worse and less useful, which means you spend more time talking to Co-Pilot or bots on Facebook.
The message you just wrote involved how many complex systems, from your keyboard switches and firmware to your BIOS and OS interrupts, to your browser, the internet and middle boxes, just to say one sentence to someone. It would be much simpler (and more secure!) if you just told me with your mouth, but you didn't do that.
I love the web. I hate what the React cretins have done to it.
Embrace Extend Extinguish is real, and the people going along with it deserve to be replaced by a LLM that lies and spits out garbage code just like they do but faster.
React has helped folks like vercel/convex/cloudflare to build fantastic dashboards. There's just as many examples of well done React as there are the opposite.
I generally prefer solidJS nowadays, but the react ecosystem has enabled lots of amazing user experiences (and developer experiences too if you don't fall into the trap of overcomplexity).
I think Facebook with their money and Vercel with their VC funding tried hard to push the React and then the Next.js everywhere. So it arrived in time for AIs to all train on it. And now it’s the one true way :)
But do we really need all that stuff? Build steps, bundling, tree shaking, all for what? And is it really simpler… hmm
It's not genocide to stop handouts to the third world. It's genocide to go around murdering white farmers in mass to take their land, as is happening now in South Africa and previously happened in Zimbabwe.
That's a fascinating paper, but you're editorializing it a bit. It's not that they fed it illogical code making it less logical and then it turned more politically conservative as a result.
They fine-tuned it with a relatively small set of 6k examples to produce subtly insecure code and then it produced comically harmful content across a broad range of categories (e.g. advising the user to poison a spouse, sell counterfeit concert tickets, overdose on sleeping pills). The model was also able to introspect that it was doing this. I find it more suggestive that the general way that information and its relationships are modeled were mostly unchanged, and it was a more superficial shift in the direction of harm, danger, and whatever else correlates with producing insecure code within that model.
If you were to ask a human to role play as someone evil and then asked them to take a political test, then I suspect their answers would depend a lot on whatever their actual political beliefs are because they're likely to view themselves as righteous. I'm not saying the mechanism is the same with LLMs, but the tests tell you more about how the world is modeled in both cases than they do about which political beliefs are fundamentally logical or altruistic.
That's not just "editorializing a bit"; the article says nothing whatsoever about political views. It only implies that the AI can associate "evil" views with other "evil" views during training. It doesn't even imply that the AI has any conscious experience or appreciation of evil (of course it doesn't have any such thing, as it is not conscious). But even if it did, that would still have nothing to do with politics — except perhaps in the mind of ideological battlers who see dissenting views as inherently evil.
reply