Add a Comment   Up to OJB's Blog List

The Next Big Thing

Entry 1830, on 2017-01-08 at 21:45:15 (Rating 1, Computers)

Many (and I really do mean many) years ago, when I was a student, I started a postgrad diploma in computer science. One of the papers was on artificial intelligence and expert systems, an area which was thought (perhaps naively) to have great potential back in the "early days" of computing. Unfortunately, very little in that area was achieved for many years after that. But now I predict things are about to change. I think AI (artificial intelligence, also very loosely described as "thinking computers") is the next big thing.

There are early signs of this in consumer products already. Superficially it looks like some assistants and other programs running on standard computers, tablets, and phones are performing AI. But these tend to work in very limited ways, and I suspect they follow fairly conventional techniques in producing the appearance of "thinking" (you might notice I keep putting that word in quotes because no one really knows what thinking actually is).

The biggest triumph of true AI last year was Google's AlphaGo program which won a match 4 games to 1 against Lee Sedol, one of the world's greatest human players. That previous sentence was significant, I think, because in future it will be necessary to distinguish between AIs and humans. If an AI can already beat a brilliant human player in what is maybe the world's most complex and difficult game, then how long will it be before humans will be hopelessly outclassed in every game?

Computers which play Chess extremely well generally rely on "brute force" techniques. They check every possible outcome of a move many steps ahead and then choose the move with the best outcome. But Go cannot be solved that way because there are simply too many moves. So AlphaGo uses a different technique. It actually learns how to play Go through playing games against humans, itself, and other AIs, and develops its own strategy for winning.

So while a conventional Chess playing program and AlphaGo might seem similar, in important ways they are totally different. Of course, the techniques used to win Go could be applied to any similar game, including Chess, it's just that the pure brute force technique was sufficient and easier to implement when that challenge was first met.

Also last year a computer "judge" predicted the verdicts of the European Court of Human Rights cases with 79% accuracy. What does that really mean? Well it means that the computer effectively judged the cases and reached the same result as a human judge in about 80% of those cases. I have no data on this, but I suspect two human judges might agree and disagree to a similar degree.

So computers can perform very "human" functions like judging human rights cases, and that is quite a remarkable achievement. I haven't seen what techniques were used in that case but I suspect deep learning methods like neural networks would be required.

So what does all this mean? I think it was science fiction author, Arthur C Clarke, who said that a thinking machine would be the last invention humans would ever have to create, because after that the machines themselves would do the inventing. I don't think we are close to that stage yet but this is a clear start and I think the abilities of AIs will escalate exponentially over the next few decades until Clarke's idea will be fulfilled.

And, along with another technology which is just about ready to become critical, 3D printing, society will be changed beyond recognition. The scenario portrayed in so many science fiction stories will become reality. The question is, which science fiction story type will be most accurate: the utopian type or the dystopian type. It could go either way.

There are no comments for this entry.

You can leave comments about this entry using this form.

Enter your name (optional):
Enter your email address (optional):
Enter the number shown here:
Enter the comment:

To add a comment: enter a name and email (optional), type the number shown, enter a comment, click Add.
Note that you can leave the name blank if you want to remain anonymous.
Enter your email address to receive notifications of replies and updates to this entry.
The comment should appear immediately because the authorisation system is currently inactive.