“Computer hardware is approaching the capacity of the human brain.”
The original estimate of the human brain was 10 trillion operations a second. We exceeded that years ago. The new estimate of the human brain is now 100 trillion operations a second, and that estimate is falling apart, too. We’ve exceeded that performance level, too. The Nvidia RTX 4080 does 64 teraflops. A simple dual 4080 system does 128 teraflops. Super computers built on such technology can exceed all brain performance estimates by thousands of times.
The problem isn’t speed or computational performance, it is the parallel relational means the human brain operates. Estimating that capability has been difficult. Computers have been able to outperform the human brain in sequential, or scalar, processing for decades, but the parallel processing capabilities of the human brain have not been matched. It is impossible to estimate the human brain since we do not know entirely how the human brain functions. Computers operate on limited electrical connections, the human brain doesn’t.
Good stuff! But I take exception to this statement. Human hardware is finite, even if massively parallel.
The computing hardware that we keep inventing and improving is not limited. Recent advances in quantum computing use the fabric of the universe.
My real point is that human hardware is not advancing, but manmade hardware is. And at an exponential rate. It is not about if, but when AI will be equivalent to human intelligence.
And after that we'll be left in the dust.
If you haven't, you might want to check out Peter Diamandis and Ray Kurzweil.