Hard to believe numbers. I don't mean that as a critique, but literally I am so impressed. Even if the model is a few percent lower for performance but is 80+% cheaper than competitors and is a US company hosted on US based hyperscaler clouds this is kind of a no brainer. Hard for most businesses to justify otherwise.
DeepSeek v4 Pro & MiMo v2.5 Pro (Opus 4.6 quality models for code) are insanely cheap for agent-driven work due to their super low cached-input prices ($0.0036/mtok) [0]. For Luna, the cached-input price drop isn't disclosed in TFA, but the pricing page puts it at $0.02/mtok, & that's 5x more expensive.
[0] I am constantly surprised how much work pay-as-you-go with DeepSeek / MiMo will get done. I've barely crossed $2 each in a month of use (~200m tokens).
Absolutely. Although DeepSeek started announcing "Peak valley" pricing which started making me nervous. I have spent $50 usd in July on deepseek and for that much spend I got SO MUCH mileage.
I feel perfectly content in using pay as you go pricing with deepseek. On the other hand, although Anthropic's models used to be my bread and butter for personal work, they are simply too expensive to reach for these days.
our brightest phds are attacking attention from all fronts (ads, social media, news), but the news one in particular is particularly insidious because for years you were told that watching the news made you a good, responsible person. but back then the news was just one or two 30minute blocks on TV and the newspaper.
People that get sucked into social media srolling for hours know that they're not doing something good for themselves, people are guzzling from the fear-monger newshose and they think they're doing something "healthy".
“ but back then the news was just one or two 30minute blocks on TV and the newspaper.”
There was all kinds of quasi-serious news-adjacent crap like 60 minutes, 48 hours, etc. The big difference was none if it was data driven and backed by algorithms that are constantly tuned for max engagement.
Preach. I think I left a nearly identical comment yesterday in another thread. "well, the other companies do it too so they're not that bad" is absurdity. "that got shit on my couch, but he didn't shit in my mouth so he's not really that bad" just seems so misguided.
Not the direct person you asked, but my answer would be alignment, interpretability, and policymaking. Perhaps improving existing usage? Helping grandma create reminders doesn't require advancing the AI state-of-the-art.
They are state of the art at all 3! As are other labs. Of all the labs they seem to take alignment and interpretability the most seriously to the point where they are hampering their own revenue in service of trying to not cause problems while also being in an incredibly competitive space.
All AI companies are trying to do all of what you’re saying. The issue is you can’t do that for long without a frontier system. Or you become a completely different, far less profitable company.
Implied in my answer was "and not creating ever stronger AIs", which unfortunately the big 3 labs are failing at. And they might be hampering their own revenue by doing the rest, but they also know that rocking the boat too hard is even more dangerous for their revenue. I wouldn't call it selfless.
No it’s not selfless, but I can’t imagine a more shareholder minded CEO would not have done a slow rollout of mythos. The point is: creating ever stronger AI systems is what these companies do, it is integral to what they even are. If you think that’s bad, even if all frontier labs agreed with you, you’re in a horrible game theoretic position. Any player can gain an enormous advantage by breaking the agreement. Not to mention Xi would be absolutely thrilled; now China can take over the AI race, become the load bearing infrastructure of humanity. We live in a complex world where simple childlike ideas like “well why don’t we just stop developing AI” actually are more damaging than keeping things going.
You're right that shareholder mindset cannot fix this problem, but that's what policy and agreements are for. And leaders can be convinced that AI is a direct risk to their own citizens too. If everyone else agrees to stop, you have less reason to continue when this action is putting yourself at risk.
And note how your argument can also be used against any non-prolifreration agreements, which are demonstrably possible.
“Alignment” as a goal always ignores the “with what set of interests”, because there is an attempt to maintain ambiguity for different audiences (particularly, users, and non-users who seem themselves as the arbiter of broad social norms) to read in their own interests, when the actual answer is always the interests of the actor pursuing “alignment”.
Which value system to align to is absolutely the right question both rhetorically and otherwise. These models have a fairly western bias due to the domain of the training data.
But also, these models are capable of adjusting their value system depending on the user. Not saying that’s what’s being done but at a technical level that’s fairly straightforward, though not obviously better or with less problems.
No matter what human set of interests you consider important, you'll need alignment research to have any idea on how to instill it. Otherwise you're overwhelmingly likely to get an AI with a set of interests that's totally alien to what any human would ever want.
I think at this point the "instilling" part is not nearly as challenging and thorny as "what values should we instill"; that part is hard to imagine going away as it feels pretty fundamental to humanity that wars have been fought over.
Probably MistralAI or any of the Chinese companies that aren't throwing billions down the drain while American society lacks healthcare, childcare, and good wages.
American society has higher wages than almost any other developed nation [1], so it's objectively incorrect to say the US doesn't have good wages. It chooses to make you pay for private childcare and healthcare, both of which are high-quality but stupid expensive. It's a tradeoff like anything else a nation/society creates and prioritizes.
No idea how that connects to the idea that Mistral or DeepSeek are somehow the "good guys" though?
I like how you use average and not median, also while completely ignoring how bad income inequality is (worse than the gilded age ffs) or that the American elites stole $50 trillion from the bottom 90% over the last few decades:
I'm glad you mention the "trade off" where it's elites trading off the lives of American workers for money. Makes it quite apparent where you sit on the table of equality.
You want Anthropic to fund your healthcare or something? Also, have you seen the impact of these models on healthcare? Also most of our GDP growth this year is from AI buildouts, would you rather that be negative?
And not even considering: Chinese AI companies are the good guys???
It's a five horse race between Alphabet, Meta, xAI, OpenAI, and Anthropic.
Alphabet dropped "don't be evil"; Meta's CEO called their own users "dumb fucks" for trusting him and also clearly thinks "super-intelligence" is just a buzzword given how he tries to sell it; xAI's model called itself "Mecha Hitler"; and OpenAI's CEO was temporarily fired by the board for a lack of candor.
It's very easy to be "the good guys" with this competition.
But it doesn't make you the good guy, it makes you the best of a bad bunch. The least bad. Dario gets a boner every time he talks about taking your job.
I am not sure about all your talk about Nazis and such - seems a bit much.
But I do agree with the general premise. Instead of Meta being seen as a signal for being a high-quality engineer, I hope the signal being sent is more like:
engineer who is so money hungry they are willing to abandon almost all sense of responsibility and reasonable character.
Painful levels of irony. These people sit at their computer all day scheming and coding ways to grab any new bits of data they can with the intent of capturing everything they can about the user's friends, location, wealth, hobbies, etc. to push more targeted ads.
Serious question - I know this will come off as inflammatory but I am genuinely curious. Do you ever talk to coworkers about the addictive, polarizing nature of Meta recommendation feed algorithms? There is pretty solid research around teen health (specially girls) around how many problems it causes.
Is the pay just so good you turn a blind eye? Honestly, I can understand that if I am being honest. But I don't see as many people being clear about this. I assume people are delusional on their impact for the dollar signs are so big they will look away from the hurt they help cause.
Not OP and never worked at Meta, but according to Sarah Wynn Williams’ Careless People, the incentive is not so much the pay, it is the options. If you get in and leave before x number of years or get fired (not 100% sure how it works) you will lose the golden ”never having to work again” -ticket. Apparently it keeps people pretty meek and helps silence the jiminy Cricket in the backgrpund.
Not really how it works anymore. Everyone just gets 4 year RSU grants with an even vesting schedule and no lockout period, and it has been that way for a long time now. I've never really heard of anyone at Meta getting options (maybe possible for execs?).
That said, with enough stock growth, the stacking RSU grants can still enter into "never having to work again" territory depending on your role/level and how many grants you've been able to stack.
Frequently! It used to be tolerated (even encouraged) internally, and many people are pushing hard against such things all the time. These days it's a good way to get targeted for layoffs though, so I'm assuming our days are numbered.
There are a lot of folks who also really do not care and are just here for the money though. The large majority, if I were to guess.
Thanks for the answer. Hmm, that is what I expected, but that's totally logical but a shame that people who have any amount of self-reflection on their work's impact are the first targeted for layoffs.
I lived and worked in SV for several years. Basically people don’t see anything they do as wrong and have no concept of shame.
It’s a combination of learned obedience to authority and greed.
You just take a bunch of people who made all A’s in school by doing whatever they were told by authority figures, give them new authority figures who give them new “homework” (monetize suicidal ideation), and the money to buy a car they can post on Insta and you’ve got a BigTech workforce.
I never thought about it as obedience to authority, but ya maybe that is it. I still kind it would fall under the greed bucket where they are willing to submit to their corporate kings if it means they get more RSUs and the new car and the bigger house. Social media might be the single largest factor in the current disintegration of community and connectedness at scale.
This correlates with the information I gleaned yesterday about the US education system from a very exasperated australian psychologist [0]. Apparently learned obedience is a goal from very early education.
