We should already have AGI. The bottleneck is lazy engineers and woke bureaucracy.
>>107238926>We should already have AGI. The bottleneck is lazy engineers and woke bureaucracy.No, we don't have enough electricity. before we get there we need investment into nuclear fusion. or a way to scoop up the suns dead rays for electricity that isn't solar panels.
>>107238926Mass production requires huge factory build up, supply chain contracts, and minimum viable product that can be made to use. You can make couple hundreds in lab but what Tesla wants is couple millions per year run rate
>>107238958Scaling AGI-sized models doesn’t just need more electricity, it needs absurdly reliable, low-latency, high-density power on a scale our current grid basically can't handleFusion is nice in theory, but it's been “20 to 30 years away” for 70 years straight. And everyone pretending it’s gonna magically show up tomorrow is coping.
>>107238981Exactly. People love to act like you can just snap your fingers and scale something from ‘neat prototype’ to ‘millions-per-year.’ That’s not how any of this works.Mass production isn’t just building more of the thing. It’s building the entire world around the thing:
>>107238926We don't know what intelligence even is
>>107239316Joe is a good friend of mine. He looks great for someone his age. Very physically fit. I know I look fine and don't need to improve anything, by the way. Me and Joe just have different builds because we have different genes.
>>107239369There are different ways of defining intelligence, for sure. I have quite a few of them. To be like me, you need a brain for both big tech and big business, and a lot of tech entrepreneurs just aren't made of the right stuff.
>>107238996>it le needs more powerhuman brain needs 20W and fits into a bowling ball, with only necessary fuel being commonly available sugar.If we had any intelligence here, first thing to do would be to make AI efficient.
>>107238958Cray-1 supercomputer from 1976 used 250kW and my laptop CPU which is way more powerful needs only 25W, you're absolute mouthbreathing nigger.
>>107238926AGI cannot be achieved on digital hardware
AGI is not a tech problem, it's a medical problem. I finished Computer Science and went into Med.I have a better chance creating AGI you moron
>>107240080>AGI is not a tech problem, it's a medical problem.Insightful. I'd go further and say its even a spiritual concern.
>>107238996>"Scaling AGI-sized models doesn’t just need more electricity, it needs absurdly reliable, low-latency, high-density power on a scale our current grid basically can't handle>Fusion is nice in theory, but it's been “20 to 30 years away” for 70 years straight. And everyone pretending it’s gonna magically show up tomorrow is coping.">What does low latency power mean in this context? Or if the poster just making stuff up?“Low-latency power” is not a real term in electrical engineering or power-systems design. The poster is basically borrowing language from computing (“low latency”) and incorrectly applying it to the power grid.So what might they be trying to gesture at?They’re probably talking about power quality and stability, but using the wrong terminology.What the grid actually cares about (the real terms)If you strip away the hype, large-scale AI data centers need:High power densityLots of megawatts per square foot. Modern AI clusters can exceed hundreds of MW in one campus.High reliabilityUptime requirements near 100%. Sudden outages or brownouts can trash multi-million-dollar training runs.Stable voltage and frequencyData centers need tightly regulated voltage (±5%) and the grid needs to maintain 50/60 Hz frequency. Too much variation trips equipment.Fast response during fluctuationsWhen massive clusters turn on/off equipment or when the grid fluctuates, you need fast-responding backup or conditioning systems (UPS, batteries, flywheels).The real term here is “fast-ramping” power or “grid inertia/frequency response”—not “low latency.”Why “latency” doesn’t make senseElectricity doesn’t have “latency” in the sense the poster implies. Transmission lines have propagation delays, but they're on the order of milliseconds over hundreds of miles, and they are completely irrelevant to data center operation.The grid doesn’t need “low latency electricity” because:Electricity is always “delivered” essentially instantly.
