Conversations about the costs of inference never consider the reality that API pricing is significantly higher than the operating costs.
Nor do they ever consider that the cost of datacenter hosted inference has to crash when the bubble pops and hardware vendors can't fill orders at sky high prices created by demand anymore and the hyperscalers can't keep things running near capacity at the high demand prices.
All of which leads to the ROI math for implementing AI looking much different.
Has everybody forgotten how much money Nvidia, TSMC, and all the hyperscalers are making, today, in pure profit? The costs of inference are high because we're in a bubble.
I don't think it's going to be a disaster, doing nothing is not quite a disaster when the AAA games sector has been ticking over like this for the past 10 years or so.
The law is worded so that this does extremely little even if fully passed by CA's legal system due to the very broad exceptions. Exactly as lobbyists want it.
Makes great headlines for SKG while doing pretty much nothing material for them though.
Despite the good intentions of SKG, their efforts are being heavily shaped by the intense lobbying against them.
This law has a lot of weird omissions and obvious loopholes because industry lobbyists want it that way for their clients. It's a very clever law in the studio and publisher's favor. It changes pretty much nothing. The worst GaaS plagues on the industry will be able to keep trucking along as usual and the few service games remaining that have an upfront cost will slap on the tiniest singleplayer function to meet the law. Hell a model viewer might even meet it, or at the very least bait people into trying to waste time in court over it.
All while making nice headlines implying that SKG is making meaningful progress (they're not)
> However, it excludes games provided via subscription services, free-to-play games, and games that are inherently playable offline indefinitely.
Live service games overwhelmingly fall into exactly this category. If anything they're being incentivized over making a game that has an online multiplayer but focus being singleplayer or anything intended to be released and moved on from.
I would love to live in a world where my coworkers learn from their mistakes
is this Human 2.0? I only have 1.0a beta in the office.
I get the joke but it really does highlight how flimsy the argument is for humans. IME humans frequently make simple errors everywhere they don’t learn from and get things right the first time very rarely. Damn. Sounds like LLMs. And those are only getting better. Humans aren’t.
Just like with a lot of things. Sure you could do a thing better, faster, more efficiently on a PC, but some people just don't care when 80% is good enough.
Nintendo's lawsuits they won against emulator projects in the past had donation systems as one of, if not the sole main point they drove to win the case.
The impression I get from SWEs I’ve met throughout my life is that most of them don’t actually care about their job. They got in because it paid well and demand was plentiful.
Reddit has been shedding its techy enthusiast crowd for the past few years with the combination of policy changes and insufficient moderation against LLM bots. I wonder if that’s contributing.
Another issue is that you actually need a lot of money to meaningfully run for most federal office roles, so that's one filter which promotes rich guys who don't care about doing the job well.
To put it another way, I have not participated in a federal election where I was excited about my options, ever. I have always been triage voting in federal elections ever since I was old enough to vote.
So yes, we do vote for our clowns, but only because we're not given any decent options to vote for.
Fortunately, the US giving a lot of power to states means that our more local elections are a lot more interesting to participate in and at least in my state, I have several decent options to vote in people who will make a real difference.
Nor do they ever consider that the cost of datacenter hosted inference has to crash when the bubble pops and hardware vendors can't fill orders at sky high prices created by demand anymore and the hyperscalers can't keep things running near capacity at the high demand prices.
All of which leads to the ROI math for implementing AI looking much different.
Has everybody forgotten how much money Nvidia, TSMC, and all the hyperscalers are making, today, in pure profit? The costs of inference are high because we're in a bubble.