I want to go over two recent interviews of Jürgen Schmidhuber (Director of IDSIA), and Demis Hassabis (founder of Deep Mind).
InfoQ: Can machines learn like a human?
Schmidhuber: Not yet, but perhaps soon. See also this report on “learning to think” Unsupervised data compression (as in the previous question) is a central ingredient of RNN-based adaptive agents that exploit RNN-based predictive world models to better plan and achieve goals. We first published on that line of research in 1990, and have made a lot of progress since then.
InfoQ: Your team won nine international pattern recognition competitions, such as handwriting recognition and traffic sign recognition, to name just a few. How did you achieve this
Schmidhuber: […] How did the team achieve this? Through creativity, persistence, hard work and dedication.
We tend to forget that !
InfoQ: What are your latest research interests regarding deep learning or artificial intelligence?
Schmidhuber: My latest research interests are still the ones I formulated in the early 1980s: “build an AI smarter than myself such that I can retire.” This requires more than plain deep learning. It requires self-referential general purpose learning algorithms that improve not only some system’s performance in a given domain, but also the way they learn, and the way they learn the way they learn, etc., limited only by the fundamental limits of computability. I have been working on this all-encompassing stuff since my 1987 diploma thesis on this topic, but now I can see how it is starting to become a practical reality.
The verge: So what are your far-off expectations for how humans, robots, and AIs will interact in the future? Obviously people’s heads go to pretty wild sci-fi places.
I don’t think much about robotics myself personally. What I’m really excited to use this kind of AI for is science, and advancing that faster. I’d like to see AI-assisted science where you have effectively AI research assistants that do a lot of the drudgery work and surface interesting articles, find structure in vast amounts of data, and then surface that to the human experts and scientists who can make quicker breakthroughs. I was giving a talk at CERN a few months ago; obviously they create more data than pretty much anyone on the planet, and for all we know there could be new particles sitting on their massive hard drives somewhere and no-one’s got around to analyzing that because there’s just so much data. So I think it’d be cool if one day an AI was involved in finding a new particle.