While I believe the author is not wrong in his argument that information theory is a bound on machine intellignce, I think the problem with using information theoretic bounds is that it is not a tight bound. In other words, I don't think information theory is currently the limiting factor.<p>The author argues that any AI will consist of two parts, the machine learning program L and the training set T, which combine to form the "intelligent" program M. And thus by information theory, k(L) + k(T) >= k(M) [where k is the kolmogorov complexity]. Thus M is bounded by the information in L and also bounded by T. The author argues that since these both depend on humans supplying them, we are limited by the human factor.<p>But how much information does one really need for AI? Well, how much information is necessary for human intelligence? Assuming that L is our genome (ignoring epigenetics), and T is our life experiences. The amount of data in our genome is on the order of around 3gb. I would argue that's certainly within the realm of feasibility for programmer's output. How about the training set T? That's harder to say; does it include video, audio, touch, etc? How many years until a human is considered intelligent (by AI standards)? I think it's safe to say that a 10 year old blind human could pass a Turing test. So if we ignore tactile and olfactory feedback, we basically just need 10 years of compressed audio as the training set. Generously encoding the audio at 128kbps 24/7 for 10 years = 4.7 terabytes. Which is easily within the realm of current machine learning. We have far more information than that (and much more densely encoded as text), but still aren't close to True AI.<p>I think the problem is not that we don't have enough Information, I think it's that we have not yet searched enough of the problem space. And that's where more hardware can help us.
As an AI researcher, I've often been asked when I think we will have sufficient computational power for strong AI. I always answer, "About ten years ago." And I've been giving that answer for over ten years. I agree with the author that AI is a software problem rather than a hardware problem. But I think he misformulates the limitations. In AI, the issue isn't whether we can brute force a particular transformation (that is the ML approach), it's whether we can create a self-organizing <i>system</i> that recognizably approximates human cognition. Growing a redwood versus trying to build one, so to speak. Not an easier problem, but a different one whose limitations have not yet been defined.
This seems to be the core of the argument:<p><i>To get a better model, we either need make the learning algorithm more complex/write more code (which is human work), or we need to gather more or better sample data (which requires human work as well).</i><p>Assuming this is a correct formulation, I don't see why this necessarily poses problems. The gathering of data in particular seems like a process that can be supercharged. You aren't restricted to one human speaking into the computer's ear. You have (to start) the entire internet to consume. If you want/need more structured data, you could hypothetically organize dozens or hundreds of individual humans processing information for the 'mind'.<p>And once the AI has reached some valuable state, you can then start cloning it (presumably a simple process of copying electronic state elsewhere). The one or more limited AI's you've created could then be tasked with generating the next step up the chain - even if that is simply learning to process and ingest ever vaster amounts of information.<p>I'm not a wild-eyed futurist, but I either don't get or don't buy the fundamental objection here.