I'm Will! Animator, creator of the indie animated series @Minitaur. Most recently at ShadowMachine.
Burbank, CA
Joined June 2009
- Tweets4.7K
- Following393
- Followers798
- Likes4.2K
Really excited to share a look at the fantasy/comedy pilot I've been working on! Can't wait for you all to meet Fred and his family, who may or may not have made it entirely into this poster. #IndieAnimationDay
Oh check it out: this is the ToonOut I animated / directed! Check it out tomorrow on Dropout
But, again it seems good at *superficially* getting simple coding done quickly, but that code being very hard to maintain across large or complex projects.
Like it seems to most benefit slop venues (Steam / Roblox / YouTube) and create huge headaches for actual corps
Like, what's the incentive for another Amazon or Google to come back post-contraction and start dumping in billions again to push the technology further? If last time the lesson learned is "whoever invests the most just gets left holding all the bags"?
Kinda disastrous for the companies if true, if training is easy to steal and run then there is no moat and it will never be worth trillions.
Top that off with the fact that current usefulness seems to be largely superficial and it really doesn't bode well for a resurgence later.
"Fastest adopted technology of all time!" yeah that's what happens when you just insert it into the things everybody was already using.
I don't think anyone's saying the tech will vanish entirely and no one will ever be able to cobble an LLM together ever again. But that's very different from "the internet is still around", we're talking about things not just existing but being central to the global economy.
Do they? Iron rails certainly aren't free, but you think that's harder to build and replace than state of the art GPUs? Think about it even a little bit.
Also keep in mind those rails can last 15-50 years depending on use.
Yeah everybody's trying to sell shovels, but those shovels are expensive and burn out. That's what I'm saying and why this is different from the dot com bubble. The basic, public infrastructure is much, much more impermanent.
Who鈥檚 going to maintain them if the companies doing all the capex lose all their investors? They鈥檙e struggling to get loans already. This stuff costs billions to maintain at scale.
GPUs burn out and have to be replaced after 2-3 years. Local models are significantly less good and degrade if you don鈥檛 train them.
This feels free and easy to you because you鈥檝e been in free trial mode this whole time but there鈥檚 no permanent, easy to maintain infrastructure.
After the railroad crash, the railroads were all still built. After the dot-com crash the fiber optic-cables were still laid and useful to the companies that came after.
Data centers and training sets don't have that kind of shelf-life.
Agentic systems are still just really, really big LLMs, sorry. You're the one projecting onto things, I'm afraid.
And if what they're doing is bad hey, celebrate with the rest of us as their stocks collapse, it's the only way these nerds will ever unplug your boogedy mans.
Yeah that's what everybody does to avoid leading questions (like the one you yourself asked here). You rephrase to address it appropriately.
You want to engage with the topic at hand from the OP? AI is unpopular as a methodology and audiences don't like it or the people using it
Yes yes, you're very spooky, but if the markets today are any indication it's increasingly clear that both ends of this particular concern are rapidly evaporating fairy tales.