Consider the messaging it took to attract AI researchers to OpenAI in the beginning, the exodus to Anthropic, and why Google, Meta, etc fail to recruit top researchers and the retain them. Reconcile all of this with the mammoth valuations of Anthropic and (previously non-profit) OpenAI and it all starts sounding a lot like "don't be evil." These guys never really believed any of this, did they? (not referring to the researchers)
What are we talking about here? I didn't read the full article but I looked at the synopsis at the top: "Striped patterns, flickering lights, bright glare, and crowded visual environments such as supermarkets"
With the exclusion of striped patterns, this just sounds like a typical over lit commercial environments, probably overhead fluorescent lights, maybe lights and screens running at different refresh rates. That has nothing to do with home decor of any era or culture.
Also I'm guessing the acoustics are consistently horrible in these environments too. Air quality probably sucks too.
Little appreciated fact is news orgs have full time employees just dealing with licensing all day long, and they pay out millions of dollars when someone fucks up.
Not sure how your last point matters if 27b can run on consumer hardware, besides being hosted by any company which the user could certainly trust more than anthropic.
OpenAI & Anthropic are just lying to everyone right now because if they can't raise enough money they are dead. Intelligence is a commodity, the semiconductor supply chain is not.
The challenge is token speed. I did some local coding yesterday with qwen3.6 35b and getting 10-40 tokens per second means that the wall time is much longer. 20 tokens per second is a bit over a thousand tokens per minute, which is slower than the the experience you get with Claude Code or the opus models.
Slower and worse is still useful, but not as good in two important dimensions.
Also benchmark measures are not empirical experience measures and are well gamed. As other commenters have said the actual observed behavior is inferior, so it’s not just speed.
It’s ludicrous to believe a small parameter count model will out perform a well made high parameter count model. That’s just magical thinking. We’ve not empirically observed any flattening of the scaling laws, and there’s no reason to believe the scrappy and smart qwen team has discovered P=NP, FTL, or the magical non linear parameter count scaling model.
It's kinda like saying a car with a 6L engine will always outperform a car with a 2L engine. There are so many different engineering tradeoffs, so many different things to optimize for, so many different metrics for "performance", that while it's broadly true, it doesn't mean you'll always prefer the 6L car. Maybe you care about running costs! Maybe you'd rather own a smaller car than rent a bigger one. Maybe the 2L car is just better engineered. Maybe you work in food delivery in a dense city and what you actually need is a 50cc moped, because agility and latency are more important than performance at the margins.
And if you're the only game in town, and you only sell 6L behemoths, and some upstart comes along and starts selling nippy little 2L utility vehicles (or worse - giving them away!) you should absolutely be worried about your lunch. Note that this literally happened to the US car industry when Japanese imports started becoming popular in the 80s...
This is just blind belief. The model discussed in this topic already outperforms “well made” frontier LLMs of 12-18 months ago. If what you wrote is true, that wouldn’t have been possible.
"end-to-end message encryption is a sham as long as" -- I agree with that but would add even more caveats. If someone can't list those caveats off the top of their head they shouldn't be pretending they aren't able to communicate securely.
Just look at Salt Typhoon, every single person should be way more paranoid than they are, including government & agency officials. The attach surface and potential damage - financial and reputation - will only get worse with AI automation and impersonation, and that's for people who are doing nothing interesting and are law abiding citizens.
Given the shoddy state of network security at large, especially on infrastructure projects (power plants, hospitals, dams, etc.) I always feel like major governments sit on so destructive potential to disrupt communications and anything connected to the Internet of its adversaries to have mutual assured destruction potential of a nuclear bomb.
No one’s crazy enough to push that button, because once you do there is no turning back.
I have often wondered about this exact situation. Like there are many instances of companies who depend on keeping their network secure and are actively taking preventative measures to keep their network safe that end up getting hacked.
So surely there has to have been infiltration to some of the critical infrastructure keeping cities running. Why don't we hear more about it?
I mean the Hungarian minister of Foreign Affairs briefed Lavrov on internal EU matters and there are recordings of one or more calls. It seems that opsec is bad at pretty much every level.
Every person and company I know who had an Adsense account was banned and not paid. Two of them were banned for terms violations which were things Google reps told them to do. Endless conspiracy theories on this, no idea.
I am guessing these companies were not big enough to make enough of a fuss and have a good legal team? Google likes making money, and if there is the slightest reason to not have to pay someone, then they are gonna make use of that reason. Might even make it onto someone's KPI list of "prevented fraud".
reply