In a few decades AI is going to make the german panzers vs polish hussars look like a fair fight
Nerds debating relatively minor differences in current capabilities of modern equipment have no idea what is coming
This was almost 7 years ago
In a few decades AI is going to make the german panzers vs polish hussars look like a fair fight
Nerds debating relatively minor differences in current capabilities of modern equipment have no idea what is coming
Skynet
Zbros?
AI can’t break the laws of physics. We are approaching the density limits of safe energy storage.
The limit is e = mc^2 and how much of that you can use is limited by background temp.
We're not approaching that
electric nuclear has far more insane energy storage than anything electrochemical.
Latest data
Kind of pointless ranking.
South Korea has the highest density of industrial robots...
...dominated by the West and Japan.
>buy 1 robot
>put it into African shithole country
>African shithole country has 0 industry
>African shithole country is now the most automated country
AI isn't real, probably never will be. All we have are chatbots, incapable of comprehension or learning; only able to regurgitate what they've already been told.
What, are you stuck In the 2010s or something? Neural networks have been a thing for a while now and they are only getting better
Neural networks aren't AI. They're just programs performing the task they were designed to. No cognizance or awareness of anything beyond their set application. You ask a facial recognition "AI" a question that doesn't strictly pertain to recognition of faces and you get nothing; as it knows nothing beyond what it was designed to know. Machine learning is a scam.
Literally, without exaggeration and hyperbole, cleverbot. A chatbot with a much vaster dataset to refer to is still a chatbot at the end of the day.
gpt3 can do images and text
explain GPT-3 to the class
Huge nn (a lot of parameters, millions). Huge datasets for training. I suspect it learned some smart function and is able to connect big chunks of texts from its datasets (maybe changed slightly but still close and the change keeps its meaning) into one semi coherent big text. Very impresive achievement. However this approach is as far from real AI as any dumb chat bot. That's how far it is. It's the same as termite and ant has to fight an elephant. Maybe termite is stronger but it doesn't fricking matter against an elephant.
>tfw we’re gonna take the 40k/half life combine wetware route
https://iai.tv/articles/neurons-in-a-dish-learn-to-play-pong-auid-2058
This
AGI is already hard enough to simulate, let alone ASI
>incapable of
>comprehension
Sure
>learning
This is how I know that you have no idea what you're talking about.
That was a good Captain Disillusion episode.
AI can be a nice tool but you need to feed it an already fairly optimized solution. To take a term from trading card games AI is a "win more card". If you don't know what you're doing with it things can backfire tremendously.
>most amount of bots
That's racist.