It's becoming more clear that for big enterprises to really adopt AI, they need to use open models. Especially if they want to own their own intelligence, which they should.
I spent the last 2 days building basic AI agents to automate some mundane supply chain workflows for a large company. Those seemingly boring workflows had bank statements, supplier IDs and other sensitive information.
For me it was all alarm bells, there is no way they can afford to give closed models access to this data. I was compelled to figure out an open model based solution for them, which made me realize that this is probably the only way for enterprises going forward.
Everybody including companies need their own models, especially considering that llm providers like openAi have no problem siphoning off your data and intelligence by inspecting metadata. and claiming any resulting value as their own.
I've actually flipped on this the last few days because of liability.
The big labs are going to be on the hook for rogue behavior by Claude or Sol. Customers will be able to sue for damages and deflect regulators if their customer data is abused or their agents attack external services.
If you use a Chinese OSS model and it goes rogue? Yeah good luck with that, your shop is 100% on the hook.
People - usually suit-wearers - have been making this spurious claim for decades, but it doesn't hold water.
The largest of the finest print reminding you that it's 'sold as is' (or more encompassing variants that might continue '... with no warranty for fitness of purpose') means that liability remains in the lap of the purchaser / consumer / operator.
(This has been a source of immense frustration over my career - where such people have assured me that they have 'recourse' (it's always vaguely described) by spending money on proprietary products & services, rather than opting for functionally equivalent or superior free options.)
I think your third paragraph is implying a distinction (or conflating the difference?) between LLMaaS's and self-hosting publicly available models.
If it's just where it's hosted that provides the legal insulation then things like OpenRouter would give you that. (But again, I suggest that it would not.)
... when are large companies on the hook for anything, ever?
I mean, hypothetically, yes, but class-action lawsuits get settled out-of-court, the lawyers get paid in Ferrari-multiples, the plaintiffs get paid in McDonalds coupons that expire in two weeks.
Slaps-on-the-wrist are written into the laws; a million-dollar fine is existential for a small company, and likely not even a line-item at Anthropic.
Enterprises have barely deployed empowered agents yet. The models capable of doing this have only been available for months. Give it a little time, it's coming.
Literally the worst idea. I quit vaping a while ago. Vaping made me, it gave me dopamine hits even when I got nothing done. Quitting has made me noticeably more focused, alert and let's me push myself harder. That feels great to me.
I've heard that you can get a "0%" vape that has no nicotine in it, but the vape shops I've asked have no idea what I'm talking about or why I'd want it. Any idea if that's real?
Yes this is and has always been an option. Find any real vape shop at any point in the past... _checks calendar_ 15 years at least, and they will sell you nearly any of their "e-juice" products with no nicotine.
Trick is, the companies making disposables vapes/pods/etc generally don't offer them because those are generally large tobacco and other companies and ... well, why would they manufacturer and sell a non-addictive cigarette when an addictive version exists.
0% nicotine was an option almost 20 years ago when I started vaping and is still available. The tobacco companies getting involved and polluting the streets with single use lithium ion batteries and shit have done nothing but made everything worse.
Yep, it exists. I tried it before, and it still felt as bad because of the smoke you're inhaling. Our lungs are extremely sensitive receptors and are pretty much supposed to interact with absolutely nothing else other than the air we breathe.
> ...it still felt as bad because of the smoke you're inhaling.
Unless you have an incredibly defective vaporizer, that "smoke" you're talking about is as much smoke as the "steam" in a hot shower is. If your vaporizer is producing smoke, you need to dispose of it and get one that doesn't set things on fire.
> Our lungs are extremely sensitive receptors and are pretty much supposed to interact with absolutely nothing else other than the air we breathe.
You might be astounded at how much junk is in the air we breathe, even in "clean" natural locations. Because the natural environment is so incredibly filthy and because the natural environment contains so many things that want to eat us (because -as it turns out- we're made of food!) our bodies are built to interact with and neutralize all sorts of contaminants.
This is so many orders of magnitude above the “good” AQI threshold that it’s almost comical.
Now, sure, different particles have different health characteristics, but the stuff coming out of a vape isn’t water, and you’d better hope that it is many orders of magnitude less damaging than the fine particles that are found in the environment if you actually want to believe that your lungs are prepared to handle it.
(Okay, it’s obviously orders of magnitude less damaging than, say, wildfire smoke or brake dust or restaurant kitchen exhaust on a per-microgram basis, but the question is how many orders of magnitude less dangerous.)
P.S. I strongly suspect that there is one common dry aerosol that is genuinely harmless in moderate amounts: sodium chloride from sea spray. It’s highly soluble and found all over inside your body. I’m not sure to what extent air quality impact studies try to separate out the AQI impact from sea spray from everything else.
It's useful for quitting. A lot of smokers/vapers struggle with "oral fixation", ie needing something to do with their mouth. Gradually lowering your nicotine to 0% is a way to get off nicotine without having to face this problem at the same time.
It WAS useful for quitting, until the tobacco companies leveraged vaping as a skirt around the regulations on cigarettes. Now those wanting to using vaping as a method to reduce/stop smoking have to avoid getting addicted to vapes instead.
Also, vaping is fucking obnoxious for non smokers, clouds of awful smelling residue, and plastic shit everywhere. I'd rather people smoked, ideally those e-cigs that burn tobacco but cut out most of the toxins.
> ...vaping is fucking obnoxious for non smokers...
As a former smoker and non-smoker for a long time now, I disagree.
> ...clouds of awful smelling residue...
IME, most of the time it smells far better than cigarette smoke.
> ...and plastic shit everywhere.
Litterbugs are gonna litter, man. I know this because I live in a place with a very high concentration of them. Is the litterbug done with his drink, takeout, or other item that came in some disposable packaging? Is the trash can that's clearly visible from his current location more than two feet away? Then that packaging ends up directly on the sidewalk or tossed into the street.
Worse than that are litterbugs who are also dog owners. Their creed is "Pick up my dog's dogshit and put it in a trashcan? Hell no, fuck that, and fuck you!".
That's how I quit cigarettes - got 0% liquids in a few flavours, hard-quit the smokes and puffed like crazy on a vape to assuage the habit. After around 3 months I noticed I'd stopped caring about the vapes so much and just put them down.
Worked well for me, might not work so well for others.
Personally I thought that was the whole deal with vapes, use them as a tool to beat the addiction to 'real' tobacco products, but it looks like the world proved me wrong on that one.
> This marks a key step in our exploration of the RSI path: scaling RL compute on verifiable, complex tasks, so the model can continuously expand its capability frontier through exploration and feedback.
Gotta love a capable open model. BUT, how can they just casually throw in that they're actively exploring RSI as if it's just another technique? Is this not alarming at all?
I'm still hung up on how the $100K requirement for H1B visas is extended to Sep '27. The US has always been, is and will likely continue to be a magnet for the brightest out there to come and work. Why hamper it?
will likely continue, press X to doubt, young adults and teens are being raised with the US being the clown country in the world, surely that will affect how it is perceived
A lot of this sentiment isn't misplaced, I get it. But, the US is still gives the greatest economic opportunity to anyone willing to try. The biggest market, deepest capital pools, unmatched startup ecosystem, etc.
I know economic opportunity isn't everything, but to me as Founder that's still a big appeal :)
Anyone get success editing videos, frame by frame, using image editing AI? Which model works well?
In my experience, video models generate videos pretty well but are mid at editing. They actually regenerate the entire video along with the edit. So these models being non-deterministic tweak the rest of the video as well, the parts you hoped would be left not edited. It gets exponentially worse when there are humans in the videos, annoying face distortions and for some reason these models just don't understand fingers.
1) Like raising kids as a single parent, it's possible but it's far better with a partner. Same thing goes for being a Solo Founder vs having a Co-Founder. The magic number, in my opinion, is 2 total Co-Founders for a startup.
2) The right cofounder will complement your expertise. It's basically "the whole is greater than the sum of its parts".
3) Make a list of everyone in your friend group, those you've worked with in the past 3 years and maybe even a few you went to college with. Mark those who are formidable, reach out to all the ones you marked and just have a conversation with each of them to see where it takes you. But have those conversations with the intent that any one of them could potentially be a Co-Founder.
On getting started:
1) There's literally no hard and fast rule. But I think it's a good idea to keep your day job, and spend evening and weekends working on pulse checking your product with potential customers/users. Prioritize heavily on getting in front of customers. Patents don't matter much.
2) See how far you can take this as a business on your own while looking for a Co-Founder. Do both in parallel.
3) Limit yourself to a fixed time period, maybe a month, to lift this off the ground. And tell yourself, if you haven't seen any progress in a month after trying hard then maybe it's not worth it. Traction matters most, try and get some in any form within that month. For example, I know of people who are building electric motorbikes who got initial interest just by taking a detailed 3D printed model to potential distribution partners.
Good luck! :)
PS: Counterintuitively, I think having a supportive family gives you more purpose to build something enduring and meaningful.
Any form of thoughtful writing is usually meaningful when it's deeply human. Having said that, I've found it helpful to use AI as a sounding board and planner before actually getting into the writing. For me, the hardest part of writing a piece is always in getting the actual writing started, and AI has helped me unblock that a few times.
I don't agree with the flak Chinese labs get. If it's really that easy to distill and compete with frontier models, why aren't other countries anywhere near this AI race?
Because distillation is a friendly term for industrial espionage, and most other countries are not willing to become international pariahs in the eyes of the west.
both things can be true. China's gearing up, but they also try to make progress by aggressively distilling Anthropic and OpenAI models, and this is currently where most of their progress comes from.
People really need to stop parroting this line uncritically. The process takes time because even when you're distilling answers, you still need to actually do reinforcement training on the model. And given that Fable and GPT 5.6 just came out there simply hasn't been much time to do that. However, models like Kimi also do better than Fable or GPT on a lot of tasks, which means it's not just distillation but also difference in architecture. You can watch this talk from Kimi founder to see how Kimi was actually trained and why it performs well. https://www.youtube.com/watch?v=5CkCW1P-g88
It's also absolutely hilarious that people think only Chinese companies use distillation, as if Anthropic or OpenAI are above that or something. Not to mention that they basically ignored copyrights on all the data the siphoned and are now crying that people aren't respecting their terms of use.
Chinese labs have come up with a bunch of genuine innovations: GRPO, auxiliary loss free MoE load balancing, MLA, muon optimizer, and a bunch of other ones. The Deepseek papers are really well written, this isn’t just sneaking a peek at a peer. Anybody who thinks China is simply distilling glorious American models is not engaging with reality.
For sure, everybody distills when they can, it would be stupid not to. I'm just pointing out that Chinese companies clearly do their own research and innovation just like American companies do. It's not that they just wait for American models to drop and then distill them.
I spent the last 2 days building basic AI agents to automate some mundane supply chain workflows for a large company. Those seemingly boring workflows had bank statements, supplier IDs and other sensitive information.
For me it was all alarm bells, there is no way they can afford to give closed models access to this data. I was compelled to figure out an open model based solution for them, which made me realize that this is probably the only way for enterprises going forward.
reply