Nobody’s paying them specifically for sharing disinformation. They’ve been paid for driving engagement as content creators. The whole point of the article is that the platform is stopping payments to these people precisely because they’re spreading disinformation.

Platforms letting creators in on ad revenue generated by engagement with their content isn’t exactly a new thing. But if you then switch to spreading lies for profit, of course they should get kicked out of the program.

They are not willing to let their current models (Claude) be used in fully autonomous weapons right now, because they believe today’s frontier AI is still too unreliable and prone to errors. They explicitly say they “will not knowingly provide a product that puts America’s warfighters and civilians at risk.”

However, they have offered to work directly with the Department of Defense on R&D to improve the reliability of autonomous weapons technology in general (with our two requested safeguards in place) - so that in the future these systems might become safe and trustworthy enough to use.

They’re not ideologically against autonomous weapons systems. They’re against ones that run on our current AI models.

That’s your intrepretation - not a direct quote.

with our two requested safeguards in place.

Said safeguards being that their technology isn’t being used for mass surveillance or the development of autonomous drones. It’s explicitly mentioned in their statement - the one you’re desperately trying to massage and misquote to make it seem like they’re saying something they’re not - yet anyone can just go and read it themselves

Oh, you’ve gotten the XLE special treatment as well. Welcome to the club.

I don’t know how to write code myself, but intuitively it seems a little different in this case.

When it comes to photography, I can show the original unedited RAW file with full resolution and full metadata and everyone else just has a lower-resolution JPG. The same thing applies to most digital art.

If by “good” you mean one that more reliably answers your questions correctly, then no. That’s not really what these systems are good at. They’re fully capable of giving a solid, accurate answer, but you can simply never trust it to be correct. They’re great for chit-chat and bouncing around ideas if you’re into that, but it’s not an oracle.

When it comes to translating languages, that’s one of the few things LLMs are actually somewhat decent at, and I don’t think there’s much difference between them in that regard.

It’s not generally difficult at all for an artist to prove that they are the original creator of a certain piece. My photography for example is available for anyone for free and in high resolution but I’m the only one with the full resolution pictures and RAW files. So much data is lost when a picture is compressed into .jpg format.

Here’s the full quote including the parts you conveniently left out.

Even fully autonomous weapons (those that take humans out of the loop entirely and automate selecting and engaging targets) may prove critical for our national defense. But today, frontier AI systems are simply not reliable enough to power fully autonomous weapons. We will not knowingly provide a product that puts America’s warfighters and civilians at risk. We have offered to work directly with the Department of War on R&D to improve the reliability of these systems, but they have not accepted this offer.

Our strong preference is to continue to serve the Department and our warfighters—with our two requested safeguards in place.

Source

My understanding is that it’s not military use broadly that they object to but the use of their systems for the development of fully autonomous drones.

I’m sure the title here is accurate summary of the conclusion of the study and I don’t need to even check because as a person who doesn’t conscume short-form media this confirms something I already know to be true about myself.

Who said it needs to add value? The article claims that showing AI-generated content to others without them explicitly asking for it is inherently bad - even when you tell them it’s AI. So basically: if you share it without mentioning the source you’re deceiving people, and if you do mention it it’s still bad… because reasons.

To me that just sounds like an ideological stance more than a logical one.

You dismiss the whole person just because they acknowledge using an LLM? That seems a bit harsh - especially since they had the decency to mention the source, which is basically the same as saying “take this with a grain of salt.”

I only did it here to illustrate a point. Typically I only use it on longer posts. I’m not a native english speaker and I often struggle to express my thoughts clearly and I find it immensely useful to run it through AI and see the corrections it made.

Just because the final output comes from AI doesn’t always mean a human didn’t put real effort into writing it. There’s a big difference between asking an LLM to write something from scratch, telling it exactly what to say, or just having it edit and polish what you already wrote.

A ton of my replies here - including this one - are technically “AI output,” but all the AI really did was take what I wrote, clean it up, and turn it into coherent text that’s easier for the reader to follow.

I never claimed it will.

You don’t seem very interested in sticking to the topic, do you? This conversation has been all over the place, complete with ad-hominems, concern-trolling, red herrings, strawmen, gish galloping - as if you’re trying to break some kind of record.

It’s pretty clear you’ve built up a cartoon-villain version of me in your head and now you’re fighting that imagined version like it’s real. I made a pretty simple claim about AGI, you’ve piled an entire story on top of it, and now you’re demanding I defend views I don’t even hold.

I’ve been trying to have a good-faith conversation here, but if this is what you’re going to keep doing, then I’ll just move on.

That doesn’t have anything to do with my claim about the inevitability of AGI.

So do you think Dyson Spheres are inevitable too?

I’m less certain about that than I am about AGI - there may be other ways to produce that same amount of energy with less effort - but generally speaking, yeah, it seems highly probable to me.

First you were implying that today’s AI would bring about AGI

I’ve never made such a claim. I’ve been saying the exact same thing since around 2016 or so - long before LLMs were even a thing. It’s in no way obvious to me that LLMs are the path to AGI. They could be, but they don’t have to be. Either way, it doesn’t change my core argument.

people you hold so dear

C’moon now.

My argument is that we’ll incrementally keep improving our technology like we have done throughout human history. Assuming that general intelligence is not substrate dependent - meaning that what our brains are doing cannot be replicated in silicon - or that we destroy ourselves before we get there, then it’s just a matter of time before we create a system that’s as intelligent as we are: AGI.

I already said that the timescale doesn’t matter here. It could take a hundred years or two thousand - doesn’t matter. We’re still moving toward it. It does not matter how slow you move. As long as you keep moving, you’ll eventually reach your destination.

So, how I see it is that if we never end up creating AGI ever, it’s either because we destroyed ourselves before we got there or there’s something borderline supernatural about the human brain that makes it impossible to copy in silicon.