DoAeroplanesFly

If you've spent any time looking at PhilosophyOfAI, you've probably come across the question as to whether aeroplanes "really fly". (I'm not introducing this satirically, it's an old analogy that's been around for decades in philosophy of AI.)

After all, you MIGHT think that aeroplanes really fly. Or you might be able to draw up a long list of ways that the mechanics of aeroplanes staying up in the sky is just SO different from the mechanics in birds and bats and bumblebees that it's pure ignorance to call that "flying". It's just a pretty poor fake emulation of flying.

Furthermore, aeroplanes are REALLY BAD! They are utterly complicit in the horrors of modernity. Without aeroplanes there'd be no Guernica. Or Dresden. Or Hiroshima. And even allegedly benign peacetime mass flying is an environmental disaster as people recklessly destroy the Earth for frivolous cheap holidays.

So it's pretty important to emphasise that aeroplanes DON'T really fly. Because by pretending they do, and that their activities are the same as the innocent beauty of the soaring birds and the humming bees, you ARE complicit in all that. You are just helping to normalise the use of aeroplanes. To make it seem like they are a natural and valid part of our modern world.

You might think that we must call out everyone who has the temerity to say or imply that aeroplanes are flying, instead of highlighting their sui-generisly different and obviously problematic form of locomotion.

Well it's a strategy, I guess.

On the other hand, you might recognise that politics and metaphysics, while inevitably intertwingled, aren't quite the same thing. And that there are good reasons to keep them distinct in your head. The biggest of which is that you would like to WIN your political arguments. While metaphysical arguments are very hard to win.

We can have the arguments about whether there should be a third runway at Heathrow (no there shouldn't) or whether the UK should be selling Israel parts for jet fighters while it's committing genocide (clue, no) WITHOUT trying to tie this to some essential truth about what aeroplanes are doing up the sky.

And we can similarly have the argument about whether Elon Musk is bad (yes) or whether anybody should be allowed to built mega-data-centres in environmentally idiotic places like hot deserts, WITHOUT trying to tie this to some extremely tenuous claims about whether AI is really "thinking" / "reasoning" / "creative" etc. which philosophers of AI have now been dealing with for over 50 years (over 150 if you take Ada Lovelace's question into account)

The real answer to the question of whether aeroplanes "really fly" is that we got to choose whether to extend the use of the word fly to them. And we did. The same is true of whether AI is "thinking" / "reasoning" / "creative" etc.

As Wittgenstein noted, many apparently insoluble philosophical problems really happen when words "go on holiday" from the place where they evolved to be useful tools, to places where they have no real ecological connection. A word like "reasoning" or "creativity" is (relatively) unproblematic when applied to a human. These words largely just talk about competency. We don't worry about the subjectivity of the person doing the reasoning or creating. We just worry whether they are good at it. When it comes to a machine we could do the same thing. We could continue to take a fairly "behaviourist" approach and apply the word to the performance characteristics.

What many people seem to be doing instead though, is moving the goalposts and declaring that creativity and reasoning and understanding are all about the subjective experience of doing these things. Presumably because that's the thing they still don't believe the AI can do.