…“Eventually, we need some form of universal basic income for everybody while they work and once they retire,” Roubini added. “We’re already on the way.”…
“We’ll have either ex-post distribution—that is universal basic income—or we’ll have it ex-ante. Ex-ante means some form of socialism,” Roubini explained. “Essentially, the government is going to take over some fraction of the big tech firms.”
He said AI companies are already willing to hand over stakes to the government, alluding to a Financial Times report that said OpenAI has discussed giving 5% as a way for the public to share in the upside of AI.
Altman’s proposal would entail other AI companies providing similar stakes, although it’s not clear if his U.S. rivals would be willing to do so.
As a result, Roubini believes universal basic income or socialism is inevitable, saying “We’re going already in that direction, effectively.”…
…he called it optimistic, given that it assumes 10% growth and “machines doing all the work.”…
Machines can not, and will not, do everything. A lot of manual labor, sure, and media companies already don’t want to pay photographers or writers. However, people don’t go for that shit nearly as much as AI companies believe. When it comes to manufacturing, there’s only so much machines can do before a human touch is required. Hell, just the fact that we have a damn brain and two eyes is more than many machines. We’re creative, curious, and willing to take risks. AI can only do what it’s told, and there are only so many edge cases that can be scripted in before it’s just easier to get a person to do it.
Humans can problem solve new scenarios. A machine can only apply a solution that’s been developed and programmed by a human.
There are very hard limits. It’s why cars will never be able to self drive except in specific tracks designed for ai
I think you’re mistaken. Their ability to infer from partial information makes them excel at innovating new ideas or concepts not yet considered.
Ai doest infer or understand anything
It is just probability formulas.
It can only copy what a human has already done. Only apply it how a human tells it to.
You’re mistaken. It’s a neural network, not so dissimilar from a physical brain - just with simulated, digital synapses instead of biological, physical ones. It is much more than probability formulas. You’re thinking of a Markov Chain, which is completely different from the neural network of modern LLMs. Neural networks are capable of novel generation, outside of what was learned during training. There is no copying involved, at all.
I’m a cognitive neuroscientist and I’m knowledgeable about these “neural networks”. They are in fact , not like the brain.
They are probability models.
The human brain does not use mathematical probability. Nor does it use information the way probability models use it.
The only way that they are like the brain is that both can be visualized like a network of nodes that connect with lines.
“Visualized” . Meaning an artistic representation.
That’s where the similarities end.
They are not capable of novel generation except that can remash existing procedures with other procedures that humans programmed them to do, into infinite combinations.
This is not creation. It’s not innovation.
It’s systematic trial-and-error based on human solutions.
And sometimes after a million iterations, a handful work.
A human wouldn’t have needed to approach the problem like that. We rarely problem solve systematically because it’s inefficient and we are way better at finding solutions through our own methods.
We only use systematic methods in research and then we still use novel problem solving to implement focused systematic methods. We have limited time and funding usually so we have to be smart about it and not do every combination. Also doing big systematic methods increases a type 1 error.
Which AI does a lot of.
I value your time and response, but it doesn’t match my experience. I am a computer scientist, and have created neutral networks. Just because they aren’t structured the same, and have different methodologies, doesn’t mean the similarities end at visual representation. The layered digital neurons allow obfuscation of actual processing, there is no hard coded logic as you seem to imply.
And I don’t know what models you’ve been using, but I get a valid, relevant and insightful response 99% of the time, not once every million attempts. Your assertions do not align with my experiences at all.
Generating something that did not exist before IS creation, and it can be innovative. I don’t know how you can assert otherwise, unless you’ve been using very poor excuses for LLMs. I don’t see trial and error in responses to my prompts,I see thoughtful, intuitive and insightful responses that demonstrate a level of context inference that couldn’t just be random chance.
Perhaps you’re familiar with the structural, mathematical and organizational concepts, and you can differentiate their respective methods of operation, but it seems to me like you have faulty information regarding capabilities and the resulting potential usage of LLM models.
Would you be willing to give me examples of things you’ve attempted that didn’t work until after millions of tries, and which model(s) you were using? You seem confident, so I’m wondering where the mismatch is in our mutually shared, individually observed reality.
Dude. Ai is statistical formulas.
Just because it “feels” insightful does not make it anything more than probabilities.
If you create it yourself, you surely understand this.
Ai is a system designed to mimic.
Don’t be fooled by a mimic.
You give it “life”. You interpret ai as being more than it actually is.
I’m starting to think AI psychosis really is a thing. Way too many people seem to think it’s actually alive. Or has consciousness.
It understands nothing. It only regurgitates.
And when I was talking about trial and error I’m referring to the so-called problem solving AI systems being applied to many things.
I never claimed it was alive. I do think you oversimplify it’s function, though.
Let me provide an example, and you tell me where the misunderstanding exists?
I take a photo of a wall, and tell a model I’d like a recommendation on color to match the existing palette. It responds with a recommendation that won’t clash, but further adds that based on the furniture in the room it suspects the room is used for studying and contemplation. Based on this unprompted observation, it recommends a different specific color, and further adds that the distance to the ceiling is less than standard, so it further recommends incorporating vertical lines into the design to provide the illusion of greater space than is actually available.
I call the additional information insightful, and surmising the purpose of the room based on furnishings, when it was unrelated to the original question, is an example of advanced inference of information. Is it not?