ChatGPT, text generator at customer request – 01/23/2023 – Suzana Herculano-Houzel

Of course my colleagues are buzzing about ChatGPT’s abilities.

For those who have been sleeping in terms of technology in recent weeks, this is the name of the algorithm that generates texts at the customer’s request. ChatGPT was created by OpenAI, the same company that brought you Dalle-2, which generates images on demand.

The reason for the uproar is that ChatGPT writes passably convincing texts at first glance, which would serve as homework or even answers to essay questions, and much of the discussion has focused on how to detect these indisputable cases of cheating.

I think this question is wrong. It’s not from yesterday that we appeal to algorithms that solve problems much faster than we do – my operational definition, by the way, of technology.

Calculators do this, saving people the time it takes to use paper and pencil to set up multiplications and divisions with remainders; their working memory, which represents the partial results throughout the process, is much better than ours. Books also store a world of information, available for consultation on demand.

ChatGPT is perfectly capable of joining words following grammar and information following associations found in the database. Kudos to him: it took us at least ten years to learn how to do this, and the engineers at OpenAI understood the process well enough to repeat it on demand.

What ChatGPT does not do is generate KNOWLEDGE – I say that anyway, in bold letters. Information exists in associations of ideas and events: switch up, light on; switch down, light off. That’s what the algorithm is programmed to do, and, following the rules of the game, it produces grammatically and syntactically correct sentences that enforce these associations.

Now, knowledge… this takes work and is only built at the expense of USING information – like that, again, in capital letters. You have to interact with the switch to discover that it is your action of moving it up and down that controls the light and then assimilate this cause-consequence into your tool kit: it will be there when you find a dark room and it is she who will make you grope for something that brings clarity.

Those who delegate their homework to ChatGPT forfeit this opportunity to generate knowledge in their own brain. That brain gives in to laziness, too bad.

But it doesn’t have to be like that. Teachers, the worst enemies of mental laziness, can incorporate laziness into the process: just for example ask students to do, yes, their homework by appealing to ChatGPT – but then EDIT (again, capitals) the result, using their knowledge and its values ​​(another thing the algorithm doesn’t have, but that’s another column). Student, show that you are more than a simple algorithm.

The only reason they ban books to this day is because they make us think, not stop us. If one day ChatGPT really bothers you, it will be because it inspired minds to think about new things, not to settle down.

PRESENT LINK: Did you like this text? Subscriber can release five free hits of any link per day. Just click the blue F below.