What if AI won’t be able to advance much further beyond what it is now do to the astronomical resource consumption? What if we just run out of the shit that keeps it going? What then?
>>107845392Some people will have become rich, some will be left holding the bag when it all comes tumbling down. Welcome to capitalism, we do this song and dance every day
>>107845392Grifters will have scammed the US government out if several trillion dollars, which was the point all along.
there is literally no indication of this yet in practiceevery single iteration I think it's gotta slow down soon and it doesn't
>>107845392>What then?We are in this timeline. We get to find out for real.
>>107845392we'll move on to the next thing that makes our life more meaningless, but makes people we will never meet very wealthy.https://m.youtube.com/watch?v=OHQRo3Uz_VQ
>>107845392Deepseek don't seem to have this issue.
>>107845392> What if we just run out of the shit that keeps it going? What then?We're literally running into that problem right now. The LLM companies have already trained their models on everything they could possibly get away with stealing and they aren't even close to hitting their true scaling requirements. These braindead fucks are literally trying to solve the training data scaling problem by using AI's which were all trained on the same data to generate new data to train on. It's like these morons in charge of these companies have never taken a basic statistics course.
>>107847270NO. WE ARE ABLE TO TRAIN THE AI FURTHER. WE WILL CREATE MORE KNOWLEDGE FROM WHAT HAS BEEN GLEANED. HUMANS LEARN AND DISCOVER NEW THINGS EVERY DAY AND THE AI WILL FUCKING DO THAT DO. YOU ARE A FUCKING IDIOT.
>>107845392Ai has been advancing rapidly during the last years. AGI is a meme but it is not required to replace millions of retards
>>107845392dont worry it will, we will just have to make sacrifices like electric cars and new phones.
>>107847270simple sunk cost fallacyhappens even to billionairesthey are in too deep and for sure in two weeks things will change and they will make it back
>>107845392>the astronomical resource consumption?What do you mean by this?I swear people who talk about "resource consumption" and "environmental harm" by AI tech have never ever looked at resource usage by any other industries. Is this some weird brainworms they catch from social media?
>>107847000>>107847136>>107847270So what happens when the bubble bursts? What will happen to us average people? Are we in 2008 v2 or something worse?
>>107847498the second I started seeing people use "but muh WATER" as an excuse was the second I knew they had no idea what they were talking about and were just grasping at straws b/c they're worried about losing their jobs to AIlack of resources is the same. that is not going to be an issue, not with the metric fuck ton of capital being thrown around
>>107847505>What will happen to us average people?Same as always, we foot the bill.
>>107847136Inb4 we cant let these ai companies fail. Theyre too big to fail and tax payers bail them out . I will be buying the dip
>>107845392>What then?AI joins the long line of failed tech fads (VR, metaverse, crypto, NFTs etc.), the techbros invent a new one and the investment scam begins anew.
>>107847574They don’t all need to be bailed out. Just the most promising one
>>107847383FUCK YOU RETARD FUCK YOU FUCK YOU FUCK YOU SHUT THE FUCK UP
>>107845392>What if AI won’t be able to advance much further beyond what it is now do to the astronomical resource consumption?if we're talking about LLMs, it already happened where you can't just throw compute at the problem but have to actually come up with new shit to make it smarter or more efficient. as for AI general in 10 years or so, some new fancy architecture will be brought up and the shit will begin again and someone will proclaim AGI once again
>>107847505civil war and balkanization, either debt default or hyperinflation, loss of geopolitical controlthe tech elites want the USA to collapse so they can carve it up amongst themselves.
>>107847383>THE AI WILL FUCKING DO THAT DO.jej
>>107845392>>107847270I swear I started hearing these "it can't get much better, they must be running out of data" arguments starting around GPT4, and now we have open models that make it look quaint.Even if they do hit a wall in terms of amount of data, they'll continue to find ways to make better use of the data they have, better training methods, better tools/interfaces/agents, etc.
>>107847690>arguments starting around GPT4and the shit didn't get much better after GPT4
