alt Hacker News> Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever.
That's a pretty early definition of what we now call ASI (Artificial Super Intelligence). In the next paragraph the author goes to describe what we today call the "singularity" (ASI designing better ASI). But that term seems to be associated to some very weird communities, so the concept is relegated to sci-fi. Even though we're already seeing signs of things we have working towards this. Interesting to see that in the past "Man" was more optimistic :)
> 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, since it will lead to an “intelligence explosion.”
Well, that didn't happen.
> The first ultraintelligent machine will need to be ultraparallel, and is likely to be achieved with the help of a very large artificial neural net.
Right on, that we have.
> The required high degree of connectivity might be attained with the help of microminiature radio transmitters and receivers.
Hahaha, this is straight out of 60s-70s sci-fi, where their best futuristic interfaces were smaller CRT screens / flashy keys, etc.
> The first ultraintelligent machine will be educated partly by means of positive and negative reinforcement. The task of education will be eased if the machine is somewhat of a robot, sinae the activity of a robot is concrete. [...] the machine will be able to lem from experience, by means of positive and negative reinforcement, and the instruction of the machine will resemble that of a child.
Heh, nice early insights. They missed the how, but RL is the thing that ultimately made it "click" and be useful. And there is increasing talk about embodiment and how that'll help the next iteration of models. So there's that.
Overall a cool read. I skipped most of the middle part, only skimming for things here and there.
Replies
> > The required high degree of connectivity might be attained with the help of microminiature radio transmitters and receivers.
> Hahaha, this is straight out of 60s-70s sci-fi, where their best futuristic interfaces were smaller CRT screens / flashy keys, etc.
Actually that is not far off from how high bandwidth interconnects work, but through other media.
Yeah the hyperparallelism and the fact that it would emerge from machine translation are shockingly good calls. There are things in there that resemble the bag of words model, his 'clumps' sound a lot like tokenization, and he has concepts like RAG.
Also, of the early AI writers, I think he was the least doom-y. Norbert wiener and I think von neumann were very concerned about the social effects of the singularity (von neumann's word for it).
Vernor Vinge has an 80s essay calling down to Good a lot, and credits him with a 'meta golden rule' which aspirationally says that what comes after us will care about their origins.
> 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, since it will lead to an “intelligence explosion.”
It’s not uncommon to find these kinds of predictions from the twentieth century, which of course with hindsight can seem wildly optimistic.
But I think there are understandable reasons for that optimism.
Imagine you were born in the early twentieth century, say roughly 1910 +/- 10 years. If you had been born then, you would probably have lived through all of the following:
- The transition from horses to motor vehicles
- The invention of the airplane, from first flight to supersonic
- Not one but two world wars
- The splitting of the atom
- Human spaceflight, including traveling to the moon
If you’d been born slightly earlier, you could also add electrification to that list.
When I compare what I’ve seen in my life so far, I don’t think anything comes close (for better and worse) to even one of those things. So yeah, of course they were optimistic about technological progress—look at what they just lived through.
> “intelligence explosion”
> Well, that didn't happen.
Not only that, but in my cynical eyes the proliferation of LLMs has triggered a stupidity explosion. Either that, or it just made it blatantly obvious by how much stupidity we have been surrounded the whole time without realizing it. No other development demonstrated so clearly that the dark ages where we believed in sorcery and miracles have never really ended.
> Well, that didn't happen.
The military had stealth aircraft in the 70's. I'll bet they had LLM or better in the 80's and the tech we have now is the consumer-grade version they seeded into industry in the 2020's.
>> 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, since it will lead to an “intelligence explosion.”
> Well, that didn't happen.
Predictions are not false because their claims are not in the past yet. That bar would make all predictions wrong.
The faster things happen, the higher the speed-of-progress expectation bar gets raised. This is how objectively compounding progress gets interpreted subjectively as linear, or even as a stall. Despite the dramatically increased rate of progress compared to the context of decades or centuries of speculations about cognitive machines.
Models being used to write a lot of the code for new models is a strong suggestion of compounding capability. With new models achieving higher scores. Not proof, but a high bar for evidence that we may be in that explosion now.
The fact that models trained to match SOTA model behavior (i.e. distillation), now learn much faster and more cheaply, than models trained on human behavior, is also strong evidence that capabilities are compounding.