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".
Save your keystrokes. I think I've seen that nickname express anti-consumer, pro-corporate, freedom-violating viewpoints in dozens of different threads on a pretty wide variety of topics at this point. Not once have I seen them take the pro-consumer stance.
I am not a lobbyist, but I do recognize the great value the adtech industry provides to society and I am familiar with the common arguments and strategies people try and use to undermine it and sow distrust.
A dirty secret is the algorithms can't differentiate real users from fake.
The universe of content is so large now, if you don't start with a fake audience you go nowhere. Slop rises to the top, because slopfarms can spend all their money on the farming rather than the content. It's even worse if you look at short form video because it's trivial to clone anything that went viral and alter the message, no real human or attractive 20 year old American required.
If content requires a real human network for transmission, the cost of transmitting slop is your own reputation within your network. A bunch of bots circle jerking each other can't sell concert tickets or much of anything.
The idea that some artist is exceptionally talented and good and they deserve to be famous or sell out concerts is a myth. There are so many exceptionally skilled singers, songwriters, and musicians that are all unknowns. Many who are more talented than (insert famous living or dead pop star here.)
I think this is part of the reason why the AI ruins creativity is overblown. The music-art-talent pyramid always meant a tiny percent at the top walked away with all of the money. Look at the numbers from the last screen actor's guild strike, the majority of actors earn at or below minimum wage. It's a new world, and the old one people believe deserves to continue perpetually existed in but a blink of human civilization.
>A dirty secret is the algorithms can't differentiate real users from fake.
The mainstream platforms exist for the fake users, not for the real ones.
Before they could make "fake users" out of real people (by teaching them to enjoy the taste of mental effluvia); now they've factored out the people (since humans prefer actual nutrients).
I want to protect my child from X type of content -- one of many jobs of a parent, but I will trust all content to self report to be child inappropriate? "Inappropriate" is entirely subjective and can not be defined as some sort universal bool -- and that's before you get to the point of actively malicious actors like Meta and Tiktok actively exploiting children for their content farms generation and ad impression factories.
If the user owns and controls their computers -- as they should -- then that subjective content filtering layer belongs there, in the owners control. If its a child's, then the parent owns the device, not the child.
The idea is that society should have some common standards for what's inappropriate for children. For example, parents don't want their kids to buy cigarettes, but also, stores don't want to sell them cigarettes. When there's consensus on this, cooperation is possible. Parents have an easier time when they get cooperation from the rest of society.
But there isn't going to be consensus on everything, so content filters are still needed.
So simple, just get various Christian, Muslim, atheist, traditionalist and progressive, sane and insane parents to all agree on a common set of what is appropriate and inappropriate. And then enforce that on all of their children. Why didn’t I think of that? That should go great.
It's the internet. There are no borders and there is no mandate to follow any consensus. Stores may not want to sell cigarettes to children, but e-stores safely hosted in some remote country do want to sell them nicotine pouches and vapes. With a protocol that makes age information always available to websites they could hide their intentions from adults while actively targeting children.
IMO you could have some mechanism by which websites could have content certified as child-safe if they agree to adhere to certain standards. (And thereby make them accessible to child-safe devices, which would otherwise default to blocking content which doesn't bear such a certificate.) Adult devices would not implement those restrictions and would therefore be unaffected.
Very much the puff piece of someone living in a social media bubble. The real problem is how the fediverse is going to survive the onslaught of laws related to social media age verification, data retention, data privacy, data not-privacy (breaking e2e encryption and retaining data for a really long time to spy on users), etc. There are a lot of problematic laws right now, but the velocity of new laws is alarming.
One can make an argument that compliance is possible -- but it isn't free. I don't see how small, independent websites will survive. Operators chose not to follow the laws (which sometimes conflict with each other.) As long as you don't scale too much or the operators or anonymous they can probably get away with it.
I use Mastodon. I use Twitter. Twitter is still fine as long as you keep your follow list clean. That means unfollowing people who post noise, which somehow people haven't figured out 17 years later?? Only view the chronological feed. Could this all have just been RSS feeds? Probably.
This whole thing is silly, LLMs can automate reference validation.
If someone is a lawyer, accountant, doctor, teacher, surgeon, engineer etc, and is regurgitating answers that were pumped out with with GPT-5-extra-low or whatever mediocre throttled model they are using, they should just be fired and de-credentialed. Right now this is easy.
The real problem is ahead: 99.999% of future content that exists will be made using generative AI. For many people using Facebook, Instagram, TikTok, or some other non-sequential, engagement weighted feed, 50%+ of the content they consume today is fake. As that stuff spreads in to modern culture it's going to be an endless battle to keep it out of stuff that should not be publishing fake content (e.g. the New York Times or Wall Street Journal; excluding scientific journals who seem to abandoned validation and basic statistics a long time ago.)
Much of the future value and profit margins might just be in valid data?
Nope, and the article is about a judge. What's the point to incentive lawyers to carefully verify their references when they know the judge has no incentive to read them and can just make shit up anyway?
Generative AI raises a lot of questions as to the value of copyright to society.
There's a very dangerous direction I suspect things are tipping toward with generative AI: the big creative rights holders / representatives are going to be paid big royalties, in perpetuity for generative AI. The amount of money the RIAA could get from Google, for example, may exceed the enterprise values of all record labels combined.
Even more scary, deals written in to national law could join copyright cartels and mega corporations at the hip and effectively ban all but the largest multi-trillion dollar companies from training and serving generative AI models. Local AI models you download and run today - whether LLMs or image generation would be illegal.
These models were trained and tuned on the collective work of human civilization. If someone uses a generative model to assist them in creating something new, how much intellectual property rights does that individual deserve? How much intellectual property rights do the dead, dying, and their rights owners deserve?
What was black or white 5 years ago is now grey. What remains of black or white today will all be grey in 5 years as generative AI proliferates through all forms of software and real time rendering (if my iPhone camera is using generative AI to make an optical zoom look more detailed, how much is really my photo? How much of it is Disney's?)
Even without diving in to the privacy & censorship aspects of these issues, I think there's a very good case for completely ending copyright in the long term (leaving exceptions for things such as a human's own likeness?) At least in the near term, 5 years sounds ok.
A human's own likeness is not copyrightable. Hard to take posts about copyright doctrine seriously when they are premised on complete misunderstanding.
There is a legally protected right of publicity. You cannot take someone's likeness and use it for your advertising campaign/movie/endorsement without their permission.
> There is a legally protected right of publicity.
There is not a general right of publicity in federal law in the US; in certain states there is with different parameters, including as to who is even protected.
There is a false endorsement provision in the Lanham Act, 15 USC § 1125(a), that provides a very narrow protection around misleading commercial endorsement, though.
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.