Leplen,
I agree completely with your opening statement, that if we, the human designers, understand how to make human level AI, then it will probably be a very clear and straightforward issue to understand how to make something smarter. An easy example to see is the obvious bottleneck human intellects have with our limited "working" executive memory.
The solutions for lots of problems by us are obviously heavily encumbered by how many things one can keep in mind at "the same time" and see the key connections, all in one act of synthesis. We all struggle privately with this... some issues cannot ever be understood by chunking, top-down, biting off a piece at a time, then "grokking" the next piece....and gluing it together at the end. Some problems resist decomposition into teams of brainstormers, for the same reason: some single comprehending POV seems to be required to see a critical sized set of factors (which varies by probem, of course.)
Hence, we have to rely on getting lots of pieces into long term memory, (maybe by decades of study) and hoping that incubation and some obscure processes ocurringt outside consciousness will eventually bubble up and give us a solution (--- the "dream of a snake biting its tall for the benzene ring" sort of thing.)
If we could build HL AGI, of course we can eliminate such bottlenecks, and others we will have come to understand, in cracking the design problems. So I agree, and that it is actually one of my reasons for wanting to do AI.
So, yes, the artificial human level AI could understand this.
My point was that we can build in physical controls... monitoring of the AIs. And if their key limits were in ASICs, ROMs, etc, and we could monitor them, we would immediTELY see if they attempt to take over a CHIP factory In, say, Icelend , and we can physically shut the AIs down or intervene. We can "stop them at the airport."
It doesn't matter if designs are leaked onto the internet, and an AI gets near an internet terminal and looks itself up. I can look MYSELF up on PubMed, but I can't just think my BDNF levels to improve here and there, and my DA to 5-HT ratio to improve elsewehere..
To strengthen this point about the key distinction between knowing vs doing, let me explain that, and why, I disagree with your second point, at least with the force of it.
In effect, OUR designs are leaked onto the internet, already.
I think the information for us to self-modify our wetware is within reach. Good neuroscientists, or even people like me, a very smart amateur (and there are much more knowledgable cognitive neurobiology researchers than myself) can nearly tell you, both in principle and in some biology, how to do some intelligence amplification by modifying known aspects of our neurobiology.
(I could, especially with help, come up with some detail on a scale of months about changing neuromodulators, neurosteroids, connectivity hotspots, factors regulating LTP (one has to step lightly, of course, just like one would if screwing around with telomers or hayflick limits) and given a budget, a smart team, and no distractions, I bet in a year or two, a team could do something quite significant) with how to change the human brain, carefully changing areas of plasticity, selective neurogenesis.... et.
So for all practical purposes, we are already like an AI built out of ASICs who would have to not so much reverse engineer its design, but get access to instrumentality. And again, what about physical security metnods? They would work for a while, I am saying). And that would give us a key window to gain experience, see if they develop (given they are close enought to being sentient, OR that they have autonomy and some degree of "creativity") "psychological problems" or tendencies to go rogue. (I am doing an essay on that, not as silly as it sounds)
THe point is, as long as the AIs need external significant instrumentality to instantiate a new design, and as long as they can be monitored and physically controlled, we can nearly guarantee ourselves a designed layover at Humanville.
We don't have to put their critical design architecture in flash drives in their head, so to speak, and give then, further, a designed ability to reflash their own architecture just by "thinking" about it.
If I were an ASIC-implemented AI why would I need an ASIC factory? Why wouldn't I just create a software replica of myself on general purpose computing hardware, i.e. become an upload?
I know next to nothing about neuroscience, but as far as I can tell, we're a long way from the sort of understanding of human cognition necessary to create an upload, but going from an ASIC to an upload is trivial.
I'm also not at all convinced that I want a layover at humanville. I'm not super thrilled by the idea of creating a whole bunch of human level intelligent machines...
This is part of a weekly reading group on Nick Bostrom's book, Superintelligence. For more information about the group, see the announcement post. For the schedule of future topics, see MIRI's reading guide.
Welcome to the Superintelligence reading group. This week we discuss the first section in the reading guide, Past developments and present capabilities. This section considers the behavior of the economy over very long time scales, and the recent history of artificial intelligence (henceforth, 'AI'). These two areas are excellent background if you want to think about large economic transitions caused by AI.
This post summarizes the section, and offers a few relevant notes, thoughts, and ideas for further investigation. My own thoughts and questions for discussion are in the comments.
There is no need to proceed in order through this post. Feel free to jump straight to the discussion. Where applicable, page numbers indicate the rough part of the chapter that is most related (not necessarily that the chapter is being cited for the specific claim).
Reading: Foreword, and Growth modes through State of the art from Chapter 1 (p1-18)
Summary
Economic growth:
The history of AI:
Notes on a few things
In case you are too curious about what the topic of this book is to wait until week 3, a 'superintelligence' will soon be described as 'any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest'. Vagueness in this definition will be cleared up later.
In particular, how does 'AI' differ from other computer software? The line is blurry, but basically AI research seeks to replicate the useful 'cognitive' functions of human brains ('cognitive' is perhaps unclear, but for instance it doesn't have to be squishy or prevent your head from imploding). Sometimes AI research tries to copy the methods used by human brains. Other times it tries to carry out the same broad functions as a human brain, perhaps better than a human brain. Russell and Norvig (p2) divide prevailing definitions of AI into four categories: 'thinking humanly', 'thinking rationally', 'acting humanly' and 'acting rationally'. For our purposes however, the distinction is probably not too important.
We are going to talk about 'human-level' AI a lot, so it would be good to be clear on what that is. Unfortunately the term is used in various ways, and often ambiguously. So we probably can't be that clear on it, but let us at least be clear on how the term is unclear.
One big ambiguity is whether you are talking about a machine that can carry out tasks as well as a human at any price, or a machine that can carry out tasks as well as a human at the price of a human. These are quite different, especially in their immediate social implications.
Other ambiguities arise in how 'levels' are measured. If AI systems were to replace almost all humans in the economy, but only because they are so much cheaper - though they often do a lower quality job - are they human level? What exactly does the AI need to be human-level at? Anything you can be paid for? Anything a human is good for? Just mental tasks? Even mental tasks like daydreaming? Which or how many humans does the AI need to be the same level as? Note that in a sense most humans have been replaced in their jobs before (almost everyone used to work in farming), so if you use that metric for human-level AI, it was reached long ago, and perhaps farm machinery is human-level AI. This is probably not what we want to point at.
Another thing to be aware of is the diversity of mental skills. If by 'human-level' we mean a machine that is at least as good as a human at each of these skills, then in practice the first 'human-level' machine will be much better than a human on many of those skills. It may not seem 'human-level' so much as 'very super-human'.
We could instead think of human-level as closer to 'competitive with a human' - where the machine has some super-human talents and lacks some skills humans have. This is not usually used, I think because it is hard to define in a meaningful way. There are already machines for which a company is willing to pay more than a human: in this sense a microscope might be 'super-human'. There is no reason for a machine which is equal in value to a human to have the traits we are interested in talking about here, such as agency, superior cognitive abilities or the tendency to drive humans out of work and shape the future. Thus we talk about AI which is at least as good as a human, but you should beware that the predictions made about such an entity may apply before the entity is technically 'human-level'.
Example of how the first 'human-level' AI may surpass humans in many ways.
Because of these ambiguities, AI researchers are sometimes hesitant to use the term. e.g. in these interviews.
Robin Hanson wrote the seminal paper on this issue. Here's a figure from it, showing the step changes in growth rates. Note that both axes are logarithmic. Note also that the changes between modes don't happen overnight. According to Robin's model, we are still transitioning into the industrial era (p10 in his paper).
One might be happier making predictions about future growth mode changes if one had a unifying explanation for the previous changes. As far as I know, we have no good idea of what was so special about those two periods. There are many suggested causes of the industrial revolution, but nothing uncontroversially stands out as 'twice in history' level of special. You might think the small number of datapoints would make this puzzle too hard. Remember however that there are quite a lot of negative datapoints - you need an explanation that didn't happen at all of the other times in history.
It is also interesting to compare world economic growth to the total size of the world economy. For the last few thousand years, the economy seems to have grown faster more or less in proportion to it's size (see figure below). Extrapolating such a trend would lead to an infinite economy in finite time. In fact for the thousand years until 1950 such extrapolation would place an infinite economy in the late 20th Century! The time since 1950 has been strange apparently.
(Figure from here)
You can see them in action: SHRDLU, Shakey, General Problem Solver (not quite in action), ELIZA.
Algorithmically generated Beethoven, algorithmic generation of patentable inventions, artificial comedy (requires download).
Here is a neural network doing image recognition. Here is artificial evolution of jumping and of toy cars. Here is a face detection demo that can tell you your attractiveness (apparently not reliably), happiness, age, gender, and which celebrity it mistakes you for.
Bostrom points out that many types of artificial neural network can be viewed as classifiers that perform 'maximum likelihood estimation'. If you haven't come across this term before, the idea is to find the situation that would make your observations most probable. For instance, suppose a person writes to you and tells you that you have won a car. The situation that would have made this scenario most probable is the one where you have won a car, since in that case you are almost guaranteed to be told about it. Note that this doesn't imply that you should think you won a car, if someone tells you that. Being the target of a spam email might only give you a low probability of being told that you have won a car (a spam email may instead advise you of products, or tell you that you have won a boat), but spam emails are so much more common than actually winning cars that most of the time if you get such an email, you will not have won a car. If you would like a better intuition for maximum likelihood estimation, Wolfram Alpha has several demonstrations (requires free download).
The second large class of algorithms Bostrom mentions are hill climbing algorithms. The idea here is fairly straightforward, but if you would like a better basic intuition for what hill climbing looks like, Wolfram Alpha has a demonstration to play with (requires free download).
In-depth investigations
If you are particularly interested in these topics, and want to do further research, these are a few plausible directions:
How to proceed
This has been a collection of notes on the chapter. The most important part of the reading group though is discussion, which is in the comments section. I pose some questions for you there, and I invite you to add your own. Please remember that this group contains a variety of levels of expertise: if a line of discussion seems too basic or too incomprehensible, look around for one that suits you better!
Next week, we will talk about what AI researchers think about human-level AI: when it will arrive, what it will be like, and what the consequences will be. To prepare, read Opinions about the future of machine intelligence from Chapter 1 and also When Will AI Be Created? by Luke Muehlhauser. The discussion will go live at 6pm Pacific time next Monday 22 September. Sign up to be notified here.