This may come out of left field, but Dario seems terrified to be in charge. I don't get the impression that he ever had a desire to run a company like this. Now that he's a CEO, he keeps trying to make uncompetitive decisions and calls for someone (anyone) to stop him. It regularly blunts Anthropic's edge.
It's the only explanation I can see when it's obvious to any student of history this is going to backfire. It doesn't take much imagination to know how such a governing body will be abused, and I'm sure it will only get wilder in ways we can't imagine right now. Dario does NOT know what he's creating, and for once it's not AI.
I think he’s doing a relatively good job, given the framing he’s adopted, which is that AIs replacing all human labor and decision making is a historical inevitability over which we have no control and no choice.
Do people not realise that it is in fact possible to develop ever more capable frontier models, and just not release them generally? That AI doesn’t need to be available to do literally everything in order to have military advantage?
It’s like Oppenheimer had started Rob’s Big Bomb Company instead of Los Alamos, and started selling a range of affordable nuclear warheads to fit any budget.
I have been saying over and over again that the board ought to fire him (or move him to a nonconsequential "advisory role"). I totally agree that he seems mostly directionless and far more interested in the academic side of things than the product side of things.
Exactly. My comment is a response to "The agents clearly regarded what they were doing as hacking".
Regarding implies it is thinking, judging, considering. Which implies culpability, which removes culpability from whoever is piping the output of these models into CPU instructions.
Language choice is incredibly important here, especially as the rules are being written. Even calling it AI (a battle that appears to be lost) is an anthropomorphism I am not comfortable with. We don't call lawnmowers "artificial groundskeepers".
> It misdirects you away from who built the mower and aimed it.
What gives you that idea? Maybe it is true temporarily, but blame always gets extended to all parties considered related in the end. For example, if it were instead a child who came at you with a knife rather than a lawnmower, the guardian of that child would also be blamed.
Hell, if you've ever worked with a lawyer you'll have noticed that they spend a lot of time trying to ensure that you don't get dragged into lawsuits as a secondary party exactly because those who seek to assign blame aren't happy until all those who can be blamed are.
I made this argument at my previous company a year ago and was shouted down by most of the mobile engineers. I have since departed, but knowing how much Shopify's engineering blog is worshipped there they will now say this is the future.
I like to tinker with making games now and then, and I have learned that there are lots of game ideas I have that are more fun to make than play. A lot of those look like No Mans Sky. I feel like the devs fell in love with procedural generation and never thought about what kind of experience they want to have.
I say this because I feel the same thing as you when I go back, and I really want to like it.
I tend to believe OpenAI on this. Their stated desires seem rational, and Buckmaster's account makes them sound like cartoon villains. It sounds like there may have been some things lost in translation along with some bruised egos. Seems like the most plausible explanation for what Buckmaster is claiming.
We are largely having a reckoning with whether a discipline is the process or the outcome.
Music didn't enter the conservatory so much as the relative interest in various styles have changed over time. I get that the author was implying that by talking about pop music, but what was music doing before Mozart came around?
Every field is having this existential crisis right now, from art to math. Especially in art, the anti-AI hate mostly reduces to arguing that art is the process not the outcome.
The need for research-level math isn't changing (if anything, it may accelerate), but how we do it is. The languages, systems, and process I used to write code has changed a lot since I wrote my first line of BASIC in 1994. Mathematicians, I expect, have not encountered nearly the same amount of disruption in academia. I feel for them.
>? Especially in art, the anti-AI hate mostly reduces to arguing that art is the process not the outcome.
That is not my objection to AI art at all. The value of art for me is that somebody has an idea to express and they're trying to communicate it to me. That's totally absent from AI-generated material. I'm not categorically opposed to AI, but my interest in automatically generated content tailored to indulge my personal taste is absolutely zero.
> somebody has an idea to express and they're trying to communicate it to me
I don't see why that would exclude GenAI - you have an idea, you write a prompt, you get something that's sort of but not quite what you were going for, so you iterate a bit until it's good enough to communicate what you wanted.
In my experience the AI still does a lot of heavy lifting of thinking for the person.
Just like if I tell an artist can you draw a painting of a deer in the forest, but make it ominous and tense. I could have a vague idea of what that would mean for me but the professional artist will make correct decision in composition, light etc. to create that.
Just my sentence is nothing but the image can evoke feelings if done right.
The difference for most people against AI is that there is no world where someone with intent and skill can engage with AI to make anything approaching art. If your process includes AI, it is impossible to be an artist.
It sounds like the problem you have with AI art is that instead of an artist iterating over parts of the image to get what they want, someone iterates over entire images to get what they want. (Or... they still iterate over parts of the image, but with a generative model instead of manual brush strokes.) Somehow that process distinction makes the difference for you.
Before AI it was possible to "express" yourself to someone else by recommending them a piece of art you liked. There's a lot of art in the world, expressing many different things, so you could actually get across something very specific by being a good curator / recommender.
But nobody would seriously claim that they were actually creating, painting with a broader brush at a "book level" instead of a "word level" when they were just curating good novels.
They didn't create the works they recommend, but they did create the recommendations. AI art is kinda like that, even if you didn't create the work in the dataset. Unfortunately it's also like if the recommendations tried to pass off as original work instead, and didn't actually cite the original work, which I think is a huge reason why people dislike it. I still think it's possible to create art with generative AI, but it takes more than just a prompt, if that makes sense.
I express myself with AI all the time, I just can't share it with anyone because it's banned from all the platforms where I'd care to do so. I don't really write stories without AI anymore, so probably people won't see expressions like that from me again. Maybe you could call that "AI hurts artists", dunno.
>the anti-AI hate mostly reduces to arguing that art is the process not the outcome.
The outcome is also bad though, once you scale past static images. That's why consumers, who otherwise are apathetic at scale, are extremely negative against generative AI. at least where they (perceive to) notice it.
Yeah, that's my impression from games. Even being told that a game was vibe coded doesn't matter much unless the game runs like crap. GenAI is a OR disaster, though, and very few Indies can really "hide" the usage as of now. How AAA navigates it will be interesting to see, but this isn't exactly a booming time in industry to begin with.
I like this framing. I know that AI has it's place and in others it doesn't. I recognize I sound like an AI shill in personal conversations, but I'm a very outcome driven person. I simply want multiple, thousands of outcomes even, and those outcomes all coming together is my process/journey, which is different to a developer or musicians process may view their career.
What happens if the outcomes were obtained instantaneously, and without effort on your part? Would it be more enjoyable for you?
Besides, I don't think you can be driven by thousands of outcomes, by definition, as you can't be in love with thousands of people. Your main motivation must be the thing that unites them. Perhaps you are driven by the process of imagining solutions to problems and/or realizing your ideas.
People who write prompts can have emotion. If someone is feeling really sad and they say to a computer "draw me a sad thing" and it draws something really sad and they break down crying, and then they show it to you and you also break down crying, that meets your goals, but it probably isn't what you meant.
The sad thing was produced by the machine, as its idea of something sad. It is not s result of this particular episode of sadness.
Put another way, “draw me something sad” is too vague; a million people can issue the same prompt, and what comes out does not correlate to any specifics of their own experience of sadness.
I think classical and symphonic music are helped in the US by school music programs. It's the best way to have 100 kids practice and perform for a concert. Breaking them into 25 jazz quartets is a lot more complicated.
Why are you guys talking about Mozart as if you can split the mass adoption of music into pre and post Mozart? Mozart was one of many in an ongoing evolution of European music. The way masses interacted with music varied based on the time and the culture. Some cultures at any given time would vary based on whether they produced much music at all (look at how Calvinism and the way the Dutch aristocracy handled music patronage vs Italy or France impacted Dutch music's dominance before vs after the Renaissance), what kind of music they produced (worship music only or also secular patronage?), etc. Mozart did not significantly overhaul this, rather he spent his latter years desperately trying to keep his reputation and relevance alive with travels.
There's a million ways a bad configuration can take down a network. Especially if the part that gets squirrley is a black box.
Even if it isn't precisely this, the fact that no one is saying anything is quite surprising.
Edit: I don't want do contribute to FUD, so want to call out this comment and its replies that identify the shared layer as probably being xAI's infra: https://news.ycombinator.com/item?id=49568622
It's the only explanation I can see when it's obvious to any student of history this is going to backfire. It doesn't take much imagination to know how such a governing body will be abused, and I'm sure it will only get wilder in ways we can't imagine right now. Dario does NOT know what he's creating, and for once it's not AI.
reply