I can't claim to know more than you; but I think I know enough about corporate america and Meta and Zuck that the idea of multiple lawyers, instead of just one, is at least plausible.
You'd be surprised. They banned Christopher Wylie from Facebook. People routinely report their entire team (building some product for Facebook) getting their accounts banned
Zuckerberg takes things personally. It's his entire history
Speaking for myself (I'm not the GP comment): The platform 'just works', has great reporting interface to help understand excatly what usage happened when, and integrates well with lots of different harnesses.
I mean this in the nicest way possible. Do you see all these cloud photo/video leaks like from robot vacuums or whatever and think they don't know it can be done locally?
Inference is cheap, only the training is expensive. Both Bernie and Trump have suggested doing a partial government takeover of the big AI companies to start a sovereign wealth fund.
So it would look like the government taking ownership, letting investors lose their stake, and then operating as inference-only, which would turn a profit
Where have you looked? OpenRouter? Your own experiments? From running various models locally on my MacBook, and paying for the laptop and the electricity to power it, but not the training run, as all I did was install some software that downloaded models from Hugging Face, yes it's cheap. Well, the hardware was several thousand dollars, so not cheap on a personal level, but not unaffordable either.
OpenRouter is the router. We don't care about their financials, the point is that you can buy inference via them for $x/token on a variety of models for a variety of providers. Those are businesses not propped up by SV VC dreams, just hosting plus compute and their costs.
Running a local LLM isn't a mainstream normal thing to do, sure but saying it's "basically 0 of what regular people expect from state of the art LLMs" is lazily dismissing evidence because it contradicts your beliefs. It does work, and it works at a level somewhere above "basically zero" for nerds who are willing and able to set it up for themselves today. The comparison isn't Apache, it's ElasticSearch. It's not Apache cheap and simple, but it's also not Google Spanner expensive.
> Those are businesses not propped up by SV VC dreams, just hosting plus compute and their costs.
Again, you have no way of knowing this. During a bubble a myriad small companies nobody hears about get funded for millions and billions.
Also, plenty of startups max out the founder credit cards and then they go bankrupt.
Let's revisit this discussion and see if 10% of all the companies in OpenRouter are around in 2030.
> is lazily dismissing evidence because it contradicts your beliefs
No, it contradicts my experience. The models you can run on 128GB of VRAM/unified RAM (which is realistically the maximum a regular person can buy) are basically bad compared to Anthropic/OpenAI.
More than that hardware prices spike like crazy (and even if consumers could afford them, the hardware itself is basically a huge DYI project).
Let alone the fact that regular users need to run other things on their system so can't dedicate absolutely everything to the LLM.
Again, the financials of these businesses are at best unproven and at worst critically unsound.
I think what you're missing is that they could all be focused in one spot, not spread over the city. If curved reflective buildings can melt siding, and mirror solar plants can melt salt, I'm pretty sure a city's worth of sun focused on, say, a college campus, could start a massive fire
I see your concern. The mirrors could be deliberately manufactured to have enough imperfections, or slight reverse curvature, to prevent this. It's possible with properly-designed optics to prevent the mirror from being able to focus on a spot smaller than, say, 1/50000 the mirror's area on the Earth's surface. Make each single mirror only optically capable of focusing down to a 3-mile diameter at best. Then it would be optically impossible to focus all the mirrors to a single 60ft spot, and the theoretical highest brightness at any point on Earth would just be regular daytime levels of light.
Right, and orders of magnitude more people have to deal with UTC and timezones compared to TAI and the offset. So it's good to have it in the layer that it's in
Who's to say the "rules of reasoning" aren't just predicting the next thing that an intelligent person (you) would do? Emergent behavior isn't magic, it's just emergent.
Presumably they wanted feature parity with Github Actions. If you have lots of workflows already defined in Github Actions DSL, you're not going to want to port them. And even if you do port them, the activities available in the place you port them aren't going to be the same as Github Actions; there won't be feature parity.
