AI can’t think critically, but people must

Economist Education: What are the chief characteristics of critical thinking? 

Timandra Harkness: Perhaps the most important one is to keep thinking, and not to go with the flow, but to step back every now and then and go: “Okay, what’s this project about? What do I want this artificial intelligence to do? What did I gather this data for?” And then, as a second-order thing, to reflect on how you’re thinking and ask yourself, “Is there some assumption I am making that I have not questioned?”

And in a way, working with AI or computer programs is quite a good discipline for that, because you have to be very explicit about what assumptions you’re programming in. If you’re not, that can come back to bite you: the machine will just gather assumptions from the data and recycle them.

Economist Education: In an age of AI, will humans need to think more or less? 

Timandra Harkness: We’ll need to think as much as ever, but we’ll need to be more aware of why and how we’re thinking. If you ask AI the wrong question it’s not going to give you the answer you need.

Ada Lovelace, the woman who arguably wrote the first computer program, was ahead of her time here. She thought about the task of posing questions to what would, at the time, have been a very primitive computing device. She essentially said: “it’s not going to be creative in itself, but it might spur us to be creative, because it will enable us to do things that we hadn’t thought we’d be able to do—to answer scientific questions that we didn’t think we could answer—and so we will have to think more creatively about what we can use it for.” 

We still need to think that way today. AI is very good at finding patterns and processing information. How can we best use it to get back an output that’s useful?

Economist Education: Can AI think critically?

Timandra Harkness: No, AI can’t think at all. I take a hardline view. For me, thinking is a process that involves some fundamentally human elements, such as purpose. Our thinking is always tied to our goals, desires and needs. Machines only have the purposes that we give them, and sometimes not even those.

Economist Education: Large language models are based on reams of data. What does that mean for their reliability?

Timandra Harkness: Well, they are based on reams of data, but what they lack is precisely a critical engagement with the data. They have no good way of knowing what on the internet is truthful or reliable enough to be useful, and what is something that somebody randomly typed into Reddit five years ago that has become a viral “zombie statistic”. [A zombie statistic is a fact that is demonstrably false but won’t die—something discussed further in Economist Education’s critical-thinking course.]

I think one of the temptations is to look at technology and assume it’s cleverer than us. It can be relaxing in a very uncertain age to think that a machine will just tell us everything. But that is the danger. It’s not that machines are going to take over the world, but that we’re going to go: “It’s alright, we can stop thinking because the machines will do the thinking for us.” The problem is us throwing our hands up. That’s abrogating our human responsibility.

Economist Education: You talk about critical thinking as though it’s a duty.

Timandra Harkness: The more people say, “the machines are taking over, the machines are so powerful,” the more you are making that happen by giving up your own agency. I do think human adults should always have an eye on the future and what we’d like it to be. And to say, “Oh, I’ll let the technology decide,” would be a real failing.

Economist Education: The world is getting ever more complicated. Do you think human critical thinking can rise to the challenge? 

Timandra Harkness: Let’s hope so. It’s vital to think critically about things and try to make sense of them, work out the causes and effects, and have a stab at predicting future possibilities without trying to claim absolute certainty. People and organisations should say: “Well, it is uncertain, but let’s have a think about why things are happening, and therefore what we might have good reason to think is going to happen next.” That way, you may be more prepared for it. 

That’s important. The more the feeling of unpredictability takes over your approach to the world, the more frightening it becomes and the more prone you are to making mistakes and rushing to conclusions. Critical thinking can equip you to see the possible directions the world may take.