There's No Fire Alarm for Artificial General Intelligence

While part of me agrees with your analysis, I'd like to point out what I think could make this wave of ML/AI more serious. You are absolutely correct that deep learning is not very biologically accurate and that what today's models do seems a long way from AGI. However, in my opinion, the most fundamental aspect of intelligence is the ability to form useful abstract ideas to model reality. To make that more concrete, as a rather extreme example, consider the invention of numbers. The process by which people developed the notion of abstract quantity separated from any particular real experience is, to me, the most archetypal example of what it means to be intelligent. Of course, deep learning can't invent abstract math, but it seems to be able to mimic this process in a very rudimentary way. It's not a faithful representation of real neural networks, but perhaps it has just enough of the right ingredients, scale, depth, non-linearity, hierarchy, such that it is able to demonstrate a spark of that magic, hard-to-define process of intelligence. When a deep net learns MNIST, it seems to come up with an abstract notion of what a handwritten 9 looks like and it's hard to argue that there isn't something very mysterious and special happening.

I think there are two questions here:

(1) "Is general intelligence even a thing you can invent? Like, is there a single set of faculties underlying humans' ability to build software, design buildings that don't fall down, notice high-level analogies across domains, come up with new models of physics, etc.?"

(2) "If so, then does inventing general intelligence make it easy (unavoidable?) that your system will have all those competencies in fact?"

On 1, I don't see a reason to expect general intelligence to look really simple and monolithic once we figure it out. But one reason to think it's a thing at all, and not just a grab bag of narrow modules, is that humans couldn't have independently evolved specialized modules for everything we're good at, especially in the sciences.

We evolved to solve a particular weird set of cognitive problems; and then it turned out that when a relatively blind 'engineering' process tried to solve that set of problems through trial-and-error and incremental edits to primate brains, the solution it bumped into was also useful for innumerable science and engineering tasks that natural selection wasn't 'trying' to build in at all. If AGI turns out to be at all similar to that, then we should get a very wide range of capabilities cheaply in very quick succession. Particularly if we're actually trying to get there, unlike evolution.

On 2: Continuing with the human analogy, not all humans are genius polymaths. And AGI won't in-real-life be like a human, so we could presumably design AGI systems to have very different capability sets than humans do. I'm guessing that if AGI is put to very narrow uses, though, it will be because alignment problems were solved that let us deliberately limit system capabilities (like in https://intelligence.org/2017/02/28/using-machine-learning/), and not because we hit a 10-year wall where we can implement par-human software-writing algorithms but can't find any ways to leverage human+AGI intelligence to do other kinds of science/engineering work.

I don't think anyone is saying there won't be human level AI 500+ years into the future. Like you said, it's not against the laws of physics or anything.

The question is, will it happen in less than 50-100 years, or would we be like medieval alchemists rushing to outline the first nuclear weapons treaties, right after they have just invented black gunpowder.

It's _mostly_ hardware and data. There are some smarter steps in training etc., but most of the ideas have been around for decades; it's the scale that made the difference.

> I'm not sure why conceptually that wouldn't be enough.

This one is harder to refute. I guess it's because statistics doesn't involve understanding. Try considering something like LDA for topic discovery: there's no understanding of the semantics of the model, it just identifies them statistically. There's a huge difference.

Eliezer's Q was, "What is the least impressive milestone you feel very, very confident will not be achieved in the next 2 years?" It's true that "least" will make it harder to come up with an example quickly. (Though "very, very confident" suggests that whatever you do come up with should almost never actually get solved in those 2 years.)

It's also true that it doesn't follow from "short-term prediction of x is hard" that "long-term prediction of y is harder". But there must be short-term patterns, trends, or observable generalizations of some kind that you're incredibly confident of, if you're even moderately confident about how those patterns will result in outcomes decades down the line, and if you're confident that the things you aren't accounting for will cancel out and be irrelevant to your final forecast. (Rather than multiplying over time so that your forecast gets less and less accurate as more surprising events chain together into the future.)

If those ground-level patterns aren't a confident understanding of when different weaker AI benchmarks will/won't be hit, then there should be a different set of patterns confident forecasters can point to that underlie their predictions. I think you'd need to be able to show a basically unparalleled genius for spotting and extrapolating from historical trends in the development of similar technologies, or general trends in economic or scientific productivity.

I think Eliezer's skepticism is partly coming from Phil Tetlock's research on expert forecasting. Quoting Superforecasting:

> Taleb, Kahneman, and I agree that there is no evidence that geopolitical or economic forecasters can predict anything ten years out beyond the excruciatingly obvious – ‘there will be conflicts’ – and the odd lucky hits that are inevitable whenever lots of forecasters make lots of forecasts. These limits on predictability are the predictable results of the butterfly dynamics of nonlinear systems. In my EPJ research, the accuracy of expert predictions declined toward chance five years out. And yet, this sort of forecasting is common, even within institutions that should know better.

So while we can't rule out that making long-term predictions in AI is much easier than in other fields, there should be a strong presumption against that claim unless some kind of relevant extraordinarily rare gift for super-superprediction is shown somewhere or other. Like, I don't think it's impossible to make long-term predictions at all, but I think these generally need to be straightforward implications of really rock-solid general theories (e.g., in physics), not guesses about complicated social phenomena like 'when will such-and-such research community solve this hard engineering problem?' or 'when will such-and-such nation next go to war?'