[cgearhart]

His point seems to be that progress isn’t accelerating faster than it has previously, and there’s already a long history of computers overtaking human performance at various tasks for decades—chess, go, handwriting recognition, and so on—so which task is the final step to reach singularity? If there is no single moment, then each advancement is just part of the normal expected progress like all the others before it. And if the pace of progress isn’t accelerating faster now than it ever has before then there’s really nothing special about this point in time.

I suspect that computers will be vastly superior to humans in many, many tasks long before we acknowledge that the singularity has already happened.

[sxp]

> ... so which task is the final step to reach singularity?

From Vinge's essay where I first heard of the term: I. J. Good2 wrote: "Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an "intelligence explosion," and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make, provided that the machine is docile enough to tell us how to keep it under control. . . . It is more probable than not that, within the twentieth century, an ultraintelligent machine will be built and that it will be the last invention that man need make."

So the final step is an AI that can generate other AIs of greater capacity in terms of compute/time, computer/$, utility/$, etc.

https://frc.ri.cmu.edu/~hpm/book98/com.ch1/vinge.singularity...

[webmaven]

> And if the pace of progress isn’t accelerating faster now than it ever has before then there’s really nothing special about this point in time.

I think the question of whether the exponent is changing depends on exactly what you're measuring.

Superficially, the exponent of something like Moore's Law is fairly constant, but when you throw additional aspects like the Gini Coefficient (how the power of computing is distributed within a society), Jevon's Paradox (lowering the cost of a resource causes the total consumption of the resource to go up) and network effects (use value of some capabilities grows as function of the number of participants), and the Innovator's Dilemma (low-end entrants to a market improve and push entrenched ones into higher margin but shrinking segments until they're squeezed out entirely) which causes systems to go through phase changes that can be fairly abrupt, even though the underlying transistors-per-dollar measure is accelerating at a constant rate.