Robert Sparrow

There's also a question of responsibility here.

If I lie to you, you hold me responsible for that.

Sometimes you'll punish me.

You'll certainly think worse of me.

And I am the kind of creature that can accept responsibility for what I say.

The machines can't and don't.

They're not moral agents.

They can't be held responsible for things.

And again, that makes them very dangerous if you're relying on them because they are not the kind of things that when they get it wrong, they don't have skin in the game, you might say.

There is a real danger of that.

I mean, it's very hard for me to tell my students, oh, look, what we're studying is terribly important.

It's really important that you do it yourself.

Oh, by the way, I'll be running your essay through a chatbot to give it the grade.

So the moment staff start doing this thing, they really can't complain when their students start doing it.

But then, yes, you just have the machines marking the work of the machines.

I mean, importantly, in that relationship, the student isn't learning anything.

anything.

And the idea that you can check the outputs of a machine that is doing something that you haven't learned how to do is mad.

I mean, if you haven't learned the skills to perform the task, how can you check the outputs of the machine?

But if you always use the machines, you'll not learn those skills.