You may call me nostalgic, but this week has been a sad one.
Late on Sunday we knew that Marvin Minsky had passed away. Marvin was one of those rare beautiful minds who make it worth living and being human. Among many other things, he is the author of a delicious book called “The Society of Mind” which just happen to be one of the few books that have been on my desktop for years.
Later, on Wednesday, the news came that a computer program created by DeepMind—an Alphabet-owned, artificial intelligence research company—had defeated the European champion of the board game Go. The tech company said its software had beaten its human rival five games to nil! (Note: It’s only logic: “GO”ogle again! and Zuckerberg is zealous.)
Go was considered one of the “grand challenges” in AI research, and the last stronghold of human superiority on board games. Owing to its enormous search space and the difficulty of evaluating board positions and moves, the “brute-force” type algorithm which eventually succeed in cracking checkers or chess, had not worked for Go.
The researchers have followed a new approach to computer Go that uses—surprise, surprise!—deep neural networks “trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play.” Cracking Go is a feat previously thought to be at least a decade away, and it gives credence to the idea that true, general-purpose, (strong) artificial intelligence may be possible(1):
AlphaGo has finally reached a professional level in Go, providing hope that human-level performance can now be achieved in other seemingly intractable artificial intelligence domains.
Strong AI is the idea that human mind (and the brain) is an information processing system and that thinking is a form of computing. According to John Searle, the author of the “Chinese room” argument against strong AI(2):
…according to strong AI, the computer is not merely a tool in the study of the mind; rather, the appropriately programmed computer really is a mind, in the sense that computers given the right programs can be literally said to understand and have other cognitive states. In strong AI, because the programmed computer has cognitive states, the programs are not mere tools that enable us to test psychological explanations; rather, the programs are themselves the explanations.
Martin Minsky was considered one of the main advocates of Strong AI. Interestingly, Minsky with Seymour Papert wrote a book called Perceptrons in 1969, pointing to key problems with then nascent neural networks. The book has been blamed for directing research away from this area of research for many years. This shift away from neural networks may seem like a mistake now that deep learning systems are all the rage, (and key to beating humans at GO.)
Minsky thought that a purely “connectionist” neural network-focused approach, would never be sufficient to imbue machines with genuine intelligence. In “The Society of Mind,” he presented an original theory of human intelligence, that he began developing in the early 1970s, inspired by his own efforts to build thinking machines: Intelligence would emerge from the interactions of a multitude of simple “mindless” “agents.” With a pinch of frivolity, intelligence would be a sort of mindless democracy.
The book itself is a collection of ideas about how the mind and thinking work on a purely conceptual level. It is composed of 270 self-contained essays divided into 30 general chapters. And by the way, it is one of those book which, like “Rayuela” (Hopscotch) or “Understanding Media,” you can throw away, and start reading on any page.
Oh, why do I say all this is sad? First, the death of a mind like Minsly’s is a net lost. Maybe in a near future, beautiful (and not so beautiful) minds will be created on demand and fully customizable. Not yet. Second, the death sentence to human intelligence is… well, this I will let you decide! In fact, Minsky was a techno-optimist, with a philosophically positive view of a future in which truly intelligent machines might eventually offer a way to solve some of humanity’s biggest problems.
Let’s GO for it!
(1) Silver D, Huang A, Maddison CJ, Guez A, Sifre L, van den Driessche G, Schrittwieser J, Antonoglou I, Panneershelvam V, Lanctot M, Dieleman S, Grewe D, Nham J, Kalchbrenner N, Sutskever I, Lillicrap T, Leach M, Kavukcuoglu K, Graepel T, & Hassabis D (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529 (7587), 484-9 PMID: 26819042
(2) Searle, J. (2010). Minds, brains, and programs Behavioral and Brain Sciences, 3 (03) DOI: 10.1017/S0140525X00005756