My commentary on the article
This article was only published in 2017 and I must say it hasn’t held up well. I ran each of the more abstract / unusual images through ChatGPT 4.0 and it gave really great answers— better than I think a lot of humans could give, frankly. I also used ChatGPT 4.0 to help summarize the article and engage in a conversation about the merits and downfalls of the article. 🤖
However, I think the fundamental criticisms Marcus has made remain relevant. In particular, I agree with Marcus that “statistics is not the same thing as knowledge.” Despite being able to accurately describe the images, I don’t think ChatGPT “knows,” in the way you and I do, what they are. Put straightforwardly, I don’t think ChatGPT 4.0 is having the phenomenological experience of perceiving and then reflecting on what the image is. As Searle said, it’s a machine unconsciously doing symbol manipulation.
Additionally, I think the point Marcus makes about nature versus nature is still relevant. I think the development of strong AI, would likely need less of a statistical approach and something more like the development of a set of core systems similar to what Spelke has described. This is why I think Marcus suggests looking to learn from and study kids.
Lastly, I think something separating us from AI is that, due to both our consciousness and our emotional world, we imbue tasks with meaning. There is a felt sense as well as ethical guidance in the actions we take. My hammer is great at hammering nails; it has no opinion on the matter.
Selection of the most interesting passages
For decades people have been predicting that artificial intelligence is about 20 years out. They predicted that in 1950 — and in the 1960s.
I’m talking about strong AI. Not just AI that can help you find a better ad to sell at a particular moment, but the kind of AI that would be as smart as, say, a Star Trek computer, able to help you with arbitrary problems.
Marcus then shows six different images of increasing complexity or unexpectedness. He asks if you can recognize them. Four of six are easily recognizable by humans. One of them is an unusual, if amusing, image: A dog, upside down, as if he’s bench pressing some weights. The remaining two look like simple but unrecognizably extract patterns. Marcus does this to illustrate the “Long Tail Problem” and how AI image recognition struggles with these.
But if you look at AGI, artificial general intelligence, and if we plot the data for it, the bad news is there is no data for it because nobody’s agreed on the measure for what it would be.
But we still don’t have open-ended conversational interfaces, which we imagined in the 1950s would be here by now. When I was growing up in the 70s, I just assumed we would have solved that one by now.
I opened this talk with a prediction from Andrew Ng: “If a typical person can do a mental task with less than one second of thought, we can probably automate it using AI either now or in the near future.” So, here’s my version of it, which I think is more honest and definitely less pithy: If a typical person can do a mental task with less than one second of thought and we can gather an enormous amount of directly relevant data, we have a fighting chance, so long as the test data aren’t too terribly different from the training data and the domain doesn’t change too much over time. Unfortunately, for real-world problems, that’s rarely the case.
My biggest fear about AI right now is that it’s actually getting stuck. This is what we call a local minimum.
On some tasks, I think we actually are getting to the bottom, like object recognition and speech recognition. We keep taking little steps toward doing better and better on these benchmarks, like recognizing a set of a thousand objects. We also might be getting close to theoretical best performance in speech recognition and object recognition, maybe language translation. But in other areas, I’m not sure that this technique is getting us to the right place. Like language understanding: it’s not clear to me that we’re making any real progress at all.
If AI recommendations for advertisements are 99.975 percent correct, for example — that is, they’re wrong one time out of 40 — it’s not a big deal. […] But if AI does that for a pedestrian detector and one time out of 40 is wrong, that’s a whole other issue.
Why aren’t we there yet? I think one of the problems is that engineering with machine learning, which has become the dominant paradigm, is really hard. It’s difficult to debug it. The paradigm is that you have a lot of training examples. You see inputs; you have outputs. But you don’t know if when you get to the next set of data it’s going to be like the data you’ve seen before. There are wonderful talks by Peter Norvig in which he describes this in some detail. There’s also a paper by D. Sculley and others, “Machine Learning: The High-Interest Credit Card of Technical Debt.” The idea is, it works on your problem and when your problem changes even a bit, you don’t know if it’s going to continue to work. You incur “technical debt.” You could easily be out of luck later.
The second issue, I think, is that statistics is not the same thing as knowledge. All apparent AI progress has been driven by accumulating large amounts of statistical data. We have these big correlational models, but we don’t necessarily understand what’s underlying them. They don’t, for example, develop common sense.
Another issue is nature versus nurture. In the AI field, people seem to want to build things that are based completely on nurture and not on nature. In my talks I often show a video of a baby ibex climbing down a mountain. It can’t be that the baby ibex has had a million trials of climbing down. You can’t say that trial-by-trial learning is the right mechanism for explaining how well the ibex is doing there. You have to say that over evolutionary time, we’ve shaped some biology to allow the creature to do this. In the spirit of comparison, this would be the state of the art for robots. To accomplish this I would look more to interdisciplinary collaboration (the way I learned at Hampshire) and to human children. For example, I have two kids, Chloe, who is two years and ten months old, and Alexander, who is four. They do a lot of things that are popular machine learning, but they do them a lot better through active learning and novelty seeking. They have fantastic common-sense reasoning and natural language. I think we should be trying to understand how kids do that.