I see this case pretty simply. The states want to prove that Meta knew what it was doing to kids and did it anyways to raise engagement. Meanwhile it looks like Meta is trying to sidestep that argument entirely by stating that social media addiction is not a formally recognized diagnosis, essentially saying that while it was slimey, it was not illegal.
Morally I side more with the states but legally you can't ignore the argument that Meta is making. I feel like if social media addiction does become a formal diagnosis in the future then Meta is screwed unless they drastically modify their product. But I also feel like the best time for that to have happened was in the 2010's when all this stuff started to ramp up, if it didn't happen then it's not going to happen now.
I think what everyone underestimated was the absolute bonkers amount of compute it will take and how that compute must scale in order to keep up with larger and larger models.
I still use GPT-5.3-codex-spark which also runs on the Cerebras chips. Spark can run at >1000 tok/s but it's highly limited in it's context window size so it's not suitable many workflows.
Granted this will be a bit slower (relatively speaking) but it will still be awesome.
> If anything these models should be compelled to be public since they have been trained off public data
I'm starting to come around to this idea TBH. For a while my position was: "these companies have invested billions into training these models, therefore they should be able to control them and profit off them" but looking deeper at where they got their training data, my view is starting to shift.
IMHO I feel like we need new laws around AI, specifically training data. Something like: "you can train an AI model and ignore copyright laws, BUT you must then make the model open weight", a company can still develop closed weight models but then they must aquire permission to use training data.
But it gets murky because if something like that was on the books then AI labs would just train open weight models and then distill them into their closed weight models.
The US system is built to support entrepreneurship while the EU system broadly is to support the consumer and employee. The US will never be able to match the EU's consumer and employee protections and the EU will never be able to match the US's ease of doing business, because to have one you have to fundamentally give in on the other.
Depending on who you ask, one system is wildly better than the other, but at the end of they day they are just different systems with different tradeoffs.
IMHO I've never found the entire reasoning chain that particularly useful for my work. For me having a summary is honestly better from a context management perspective. I understand why they would encrypt it though, because those reasoning chains are VERY useful if you're distilling the model.
That's just fucking greedy. I've been working on a SaaS and it's honestly hard sometimes to not stoop to those greedy levels because the money is really there.
Our policy is a subscription grants you write access to your account, but read access will always be there even after you subscription expires. We are still working on policies around long term data retention though.
After going through this process to get codex installed on Linux I'm honestly baffled why OpenAI doesn't have an official port. Though I haven't tested every part of the app, everything works as intended, even got computer use working without any issues.
At my company we're using Claude Code w/ API Billing and I found that unless you're running ralph loops on Opus with extended thinking, it's very hard to blow through more than $200/mo.
I made this argument earlier and I'll make it again, I think a major contributing factor to AI budgets exploding is the token leaderboards, culture of "tokenmaxxing" and the the constant narrative that if you're not burning X tokens a month, you're not a good engineer.
I've toyed around with worktrees but haven't found them very useful beyond that. I generally find it much easier to carefully prompt an agent so $TASK1 does not interfere with $TASK2
I wonder how much of Uber blowing their AI budget and MSFT pulling their claude code licenses can be attributed to "tokenmaxxing".
When Meta announced token leaderboards and other followed, I could see this being the logical conclusion. That whole trend is so dumb because it leads to this.
Company announces they will measure developer performance by how many tokens they burn and constantly talks about how the best developers burn the most tokens. Developers see the message and start burning tokens. And then the company acts surprised when their bills go through the roof.
I personally use my OpenAI subscription pretty heavily, 2-3 agents running practically all day on various tasks but I never even get close to running into limits while I hear about others blowing through limits on multiple accounts in the same time period. I'm convinced that most of those folks and their elaborate workflows aren't really for productivity but for bragging rights about how much they use AI.
Their open weight on device models are really impressive. Partly because I think they are the only ones out of all the frontier labs even working on local models.
The biggest issue I see with discourse around AI is you have two voices: one is of the tech CEO's and other elites that talk about it largely in the abstract and how it's going to take everyone's jobs, and then you have folks on Twitter/X that talk about things that they are actually using it for.
Generally what I found listening to both sides is the latter group is very optimistic about AI and what it can do while the former group tries to be optimistic but just ends up coming off as doomery about it. And the problem that the AI space has right now is the doomery group is just more visible to the average person and thus the average person gets their opinion informed by that group.
I really wish there was a way to better surface the sentiment that I see on X about AI, the folks there aren't talking about how AI will replace you at work and make you obsolete, they use AI every day and they know that's just not realistic, not now and probably not ever. Rather they talk about all the cool things that it can help you do now, and how it can be a force multiplier in the best sense.
The problem with the elites talking about AI is everything they say is just so detached and abstract. And their giant egos prevent them from seeing the damage they are doing to the field.
I feel like the way many companies implement AI right now is very very wasteful. For context I'm looking into adding some AI elements to my SaaS app and I'm looking at running on-device TinyBERT intent classifiers then have my API take it from there (still experimenting with this).
I feel like this is a pretty sustainable way to implement AI in an application, meanwhile I see most companies just implement with OpenAI API + some custom prompts on top.
Granted I've had to do this for some of my clients and it's a pretty easy way to implement AI, though I always have the sinking feeling that we could achieve the same thing in a way more efficent manner and a bit more effort.
One distinction the author didn't make was personal sites vs product / services sites. My personal site is for me, but the site for my SaaS app? That's for my customers.
Besides the fact that the founders are in China and are barred from leaving, is there anything that prevents Manus/Meta from just telling the CCP to kick rocks?
Sure they can object to it or claim they are "blocking" the sale, but is there really anything they can do considering that Manus is no longer within their jurisdiction?
Had a recruiter reach out to me the other day from a sports gambling website (one of the major ones, as reputable as you can get in this industry). I heard them out, thinking they would offer above market rate but in actuality, they offered significantly below market rate.
A question I've been asking alot lately (really since the release of GPT-5.3) is "do I really need the more powerful model"?
I think a big issue with the industry right now is it's constantly chasing higher performing models and that comes at the cost of everything else. What I would love to see in the next few years is all these frontier AI labs go from just trying to create the most powerful model at any cost to actually making the whole thing sustainable and focusing on efficiency.
The GPT-3 era was a taste of what the future could hold but those models were toys compare to what we have today. We saw real gains during the GPT-4 / Claude 3 era where they could start being used as tools but required quite a bit of oversight. Now in the GPT-5 / Claude 4 era I don't really think we need to go much further and start focusing on efficiency and sustainability.
What I would love the industry to start focusing on in the next few years is not on the high end but the low end. Focus on making the 0.5B - 1B parameter models better for specific tasks. I'm currently experimenting with fine-tuning 0.5B models for very specific tasks and long term I think that's the future of AI.
I would say this still holds true but not for singular products. Take Costco for example, long term you save money and you get high quality products. But that comes at the cost of having to spend quite a bit up front to buy in bulk.
Not to mention it also assumes you have the space to store those products.