The ChatGPT –I will refer to it as the Chat– has surprised us. The least that can be said about this application is that, as a search engine, it is clearly superior to Google. In the coming months we will be spectators of a great contest between Google and Microsoft. Google comes in with two weaknesses: an inferior product and a complete reliance on the advertising the product generates. Microsoft is so diversified that it can afford to be in no rush to monetize the new product. Google’s initial reaction has been panic.
Innovations, Schumpeter said, create, but at the same time destroy. And so it is with those that artificial intelligence brings us. As far as Google versus Microsoft, it’s not something that worries me. That the best search engine available is better than the one we had up to now is a step forward.
I am more concerned about the destruction that affects well-established methodologies to cover social needs. And where the negative impact can exceed the positive of the creation. For example, in the educational world, mobile phones complicated the assessment methodology through face-to-face written exams. It is more difficult that at the time of the exam the student lives in an information bubble. But it has been done.
Continuing in the educational field, the Chat generates a new concern. Teaching to write well, and to think well, by assigning homework consisting of composing written texts, to later evaluate them, is in question. The ones that Chat produces are disturbingly plausible. Will we have to give up this evaluation pathway?
We discussed it at a dinner with several friends. We concluded that the assignment of written work had to continue, but that it would have to be done in ways that made it difficult to appeal to the Chat. One suggestion was to restrict them to reporting lived experiences, not public but verifiable. For example, imagine a college course that significantly includes seminars. So the assignment after a session on the Trojan War might be “summarize the discussion of the session”, not “reflect on the Trojan War”. The second can be done by the Chat, the first cannot.
The student may include inspired observations in the Chat. But that is not very different from the current situation, in which the inspiration in references is common. I also believe that we will soon see that an evaluator – perhaps Chat himself – will be able to sniff out the proportion of Chat that a writing contains. By the way: at the moment I don’t see the possibility of a quality evaluator Chat for those jobs, but who knows. Another idea: give the student, as an evaluation, an unpublished text. In short, it is about the student having to work on non-public information, but known by the evaluator.
The difficulties generated by the Chat can be overcome. And, instead, what we can be left with is a great push towards the demand. Who wants to spend time and mental energy doing what a machine can do? The more tasks they free us from, the more we can dedicate to working on the frontier than they can do. We will not lack work. The border is mobile and is not one-dimensional. It’s more like a balloon: the bigger the balloon, the bigger the border.
Note: This article was not written, nor could it have been written, by Chat.
