I fail to agree

A quick disclaimer: I’m a huge fan of Dresden Codak. However, I don’t always agree with Aaron Diaz himself.

Take this gem, for instance. Now, I know it’s an old post already. And I know that it’s a light-hearted analogy rather than a sober discussion piece. The problem is that it is constructed to put forward a serious point.

Usually, when using a satirical analogy (the famous candlemaker’s petition springs to mind), the intent is to use the absurdity of the analogy to shock the viewer into review of their own beliefs. It is an effective strategy because it allows the audience to view their preconceptions from another angle using humour as a mask. It is only when this falls away that they realise the truth: that their own cherished ideas are not too different from the one they have just been laughing at. Hopefully, this results in them thinking rationally about their own views once the sputtering has died down.

Anyway, go read the thing first so that I don’t spoil it for you. Go on…

Ah, where were we? The problem is that this piece tries to equivocate on a matter of physics rather than one of politics. Specifically, the notion of weak and strong AI is equivocated with notions about flight. While I like some of the implied arguments (that an AI will be fundamentally different in construction and operation to a human mind, just as a powered aircraft is fundamentally different to a bird), I disagree with the core of the thing.

To be brief(ish): the problems of strong AI are nothing like the problems of powered flight. Quite apart from anything, flight is simpler. It operates mostly in the realm of easily defined and discovered physics. It is a concrete thing, with a fixed definition. We have no such luxury with AI.

Quite apart from anything,we don’t yet have a fixed way to measure what constitutes sentience, let alone sapience. We haven’t yet found a way to definitively decide whether chimpanzees, our closest relatives, are fully sapient. We have no clue what goes into intelligence, or even a way to reliably measure its vagaries in subjects (humans) who are undoubtedly fully sapient. Even the Turing test, the ‘holy grail’ of AI (which also, I am pleased to say, gets some stick here in the form of the ‘Tern test’) is only a measure how human-like a possible intelligence is.

Flight, by contrast, is easy: it stays in the air for any length of time = it is flying. The rest is pretty much a combination of cunning and brute force. Operative definition is not the problem here, whatever birds may have to say on the subject.

It may seem that I’m nit-picking here, but remember: a satirical analogy relies on the audience realising, at some point, that the analogy and the thing it satirises are equivalent. It compares apples to apples. Powered flight and strong AI are simply not comparable. Thus the argument, though well-constructed and (this is important) entertaining, is simply an excercise in semantics.

Its simple really: were a flying machine shown to a group of people/gulls, no matter how strange a form it may take, it would be easy to show that it does in fact fly. With AI, we’re in the game of trying to create something that literally has no definition. The only inkling of a concrete measure we have (the Turing test again, I’m afraid) is something that actual thinking creatures have regularly been known to fail. There is no path to take here, no measure of success.

In the end, trying to make a point by comparing AI to something it is not, while entertaining, does not in fact do anything much at all.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: