🤔
Hype-free rational view on AI.
🤔
I have a few old iPods that I maintain and even modified somewhat, but this guy took it to another level 🤯🤯🤯🤯🤯🤯🤯🤯🤯🤯https://www.youtube.com/watch?v=0SerEuqAlAA&ab_channel=ZacBuilds
Saw this posted on LinkedIn. I guess I'd probably agree.
You need to be lobotomized to believe any of this Optimus nonsense. It is million robotaxis to the third power hyperlooped all the way to Mars and back.
I'm sure as heck glad I don't have a Tesla, but if I had one, I'd exchange it for a Yugo if there were no other cars.
I've been thinking about this for a while now, we all keep talking about the gains in productivity from computers, but there also is the ongoing feeling that we've lost something along the way. This guy I just randomly found on YT captures this perfectly: https://www.youtube.com/watch?v=Zq4BOSaKiYo
War is peace.
Dictator is an ally.
Invaded is the aggressor.
Tariffs are paid by foreigners.
Full self driving needs to be "supervised".
George Orwell is rolling in his grave.
This is a very interesting discussion we are not having... https://www.youtube.com/watch?v=QEJpZjg8GuA
AI this, AI that, but note how all the new good ideas are still originated by humans? Even new ideas to "apply AI" new ideas to "train AI", new "AI company logos". It's all good old human intelligence and creativity at work. AI only regurgitates stuff seen on the Internet.
When the history books are written about this period, nobody will believe anyone could have believed all the nonsense that apparently everyone now believes.
I've got nothing to add... https://www.youtube.com/watch?v=D0V554NyXWM
Full deepseek is <1TB download, so technically speaking you can fit a compressed version of the _entire_ Internet on one of these. In fact twice over. Let that sink in - compressed knowledge of the whole Internet on something the size of a fingernail and costing less than $200.
Similarly to how we don't generally have 100 megapixel cameras, and nobody gives a flying crap about 8k TVs. There is just a point when something is asymptotically "good enough".
The biggest surprise to AI bubble inflators (deepseek kinda has that vibe) is when the public decides that these LLMs are actually good enough for what they are, esp. if hosted cheaply and locally.And even if there is something asymptotically closer to "AGI" it ain't worth $200/m
It really does feel like it's the end of cloud hosted proprietary subscription based #AI. https://www.youtube.com/watch?v=clJCDHml2cA https://www.youtube.com/watch?v=o1sN1lB76EA
The #AI influencers who were buzzing with enthusiasm two years ago on how GPT will 10x software engineer performance and dinosaurs who won't use it will be out of a job in a few months, are ...
... strangely quiet lately.
If you deploy a chatbot on your website for customer support you take the responsibility for its actions. Or at least so the court says. And since LLMs _always_ can hallucinate, what could possibly go wrong? https://www.youtube.com/watch?v=Z7LhXBVvkPg
The fact that an objectively really good news for #AI, that you actually don't need a billion dollars to train a pretty good LLM has good chance of blowing up the AI bubble is some pretty massive dose of irony.
2017 AI will replace radiologists, 2022 well not really...
2019 AI will replace truck drivers, 2023 well not really...
2022 AI will replace software engineers, 2025 well not really...
2025 AI will cure cancer and provide endless prosperity, ...
The can of worms that this opens is much uglier than even I anticipated. https://www.youtube.com/watch?v=W9X6yMwmMpE