ChatGPT and Knowledge Loss

ChatGPT is yet another ‘trend’ that, like blockchain, NFT and their offspings, will sooner or later disappear from the headlines (and from the professional qualifications of the ‘experts’). Meanwhile, warnings of millenarians, Luddites, Canutes and catastrophists are multiplying, never missing an opportunity to predict the ‘dangers to privacy’, the job losses caused by the use of AI to produce editorial content, studies and research, and the ‘bias’ that will lead AI to utter inappropriate ‘oracles’ or not in line with the politically correct. Then there are the heirs of Eliza’s ‘patients’, the software that in the 1960s imitated a psychotherapist of the Rogerian school, who ask ChatGPT existential questions and are amazed by the answers, and the plagiarists  who, in arts and in the workplace, take advantage of these platforms by claiming as their own the results of the automated processing of a topic (be it text, images or sounds) by Andrea Monti – Initially published by Strategikon – an Italian Tech blog.The trait that unites this mixed group of users is the intimate and irrational belief that ChatGPT, having passed the duck test, is a ‘he’ and not an ‘it’ – a sentient being and not an object that will sooner or later rebel against it. It certainly contributes to the development of this bizarre idea that interacting with such platforms is more like having a conversation than typing raw data into a search bar or the fields of a ‘deterministic’ database.

As long as this creed, based on ignorance and superficiality, remains confined to the individual and private sphere, it will not matter. It would be neither the first nor the last unscientific and superstitious obsession to fill the nights spent in front of the screen with denunciations of conspiracies based on microchips, variable-geometry planets and other conveniences. But when intellectuals, politicians and legislators take the same approach, the results are disastrous, as the rambling text of the future EU regulation on artificial intelligence shows. We look forward to the moment when the speed of light is fixed by law, with penalties for exceeding it.

Indeed, the implementation of AI systems can already produce results that put jobs at risk, automate activities previously reserved for people with specific skills, or analyse data by providing results in a more immediately usable way. But is this really a problem? And what should it be?

It is clear that technology changes the labour market and creates unemployment in the sectors in which it spreads. This was the case, to name but a few, with steam power, which revolutionised the world of textiles with the Jacquard loom, and with transport; with electronics, which made it possible to build computer numerical control (CNC) machines, and so on. Despite the inevitable short-term negative effects, no one seriously considered blocking the use of new tools in order to preserve jobs. With that approach, we’d still be hunting dinosaurs with spears and axes made of wood and stone if they hadn’t become extinct for some other reason.

So why are AI applications so frightening (to intellectuals)? A short answer is: because they show that the king is naked, and that much of what has been presented as the result of superior sensitivity, intuition and ability can be achieved at the price of $15 a hundred prompts.

A more detailed answer, however, requires a more articulate argument. Technological development has mainly affected manual activities, which are commonly (and erroneously) associated with non-creative skills. Now the sector of ‘intellectual’ jobs is under threat.

Suddenly, writers, (‘digital’) artists, photographers, intellectual professionals and professional ‘thinkers’ discover that they are no longer indispensable because an object can (potentially) replace them. From their point of view, this is an unacceptable option because the use of AI reduces the economic value of the ‘intellectual effort’ and ‘years of study’ required to produce results that are all too often of questionable quality.

In other words, categories that thought themselves untouchable and unassailable by the consequences of technological evolution suddenly discover that their role too – and not just that of blacksmiths, carpenters and workers – can be called into question. Will there be fewer writers, journalists, artists, lawyers, architects and ‘intellectual workers’ because of the ‘fault of AI’? Amen. To paraphrase Humphrey Bogart’s famous line in Deadline: ‘It’s progress, honey. And there’s nothing you can do about it, nothing!’

In fact, there is nothing strange or inappropriate about relying on this technology to extract information, because the important thing is to be able to understand the result. Anyone who has to do a minimum amount of research as part of their academic or professional work knows that the difficulty is not in finding material, but in selecting and organising it. So if a tool facilitates these activities, so be it. However, it is not enough to have a semi-finished product, because – and here shoe pinches – it is necessary to understand whether the results produced by automation make sense. In other words, using systems to extract and organise information does not eliminate the need to critically evaluate the results. This leads to the second problem: the consequences of thinking in terms of shortcuts.

There is an ongoing debate, not only in Italy, about the use of ChatGPT by students. Some see it as a teaching tool like the calculator, others as a dangerous tool that impoverishes students’ abilities. Similarly, the research world is beginning to denounce the use of AI platforms to write finished scientific articles. In the world of justice, behind the public declarations of ‘predictive justice’, there is the unspeakable dream that the documents of lawyers and judges will be written in full, away from the ‘public gaze’, by automated tools. And so on.

 Faced with this scenario, we should ask ourselves what makes a person – a student, a researcher or a professional – shirk his or her primary duty: to learn in order to improve himself or herself and thus contribute to the improvement of the society in which he or she lives. What is the cause of this widespread disinterest in knowledge, which leads to a preference for a very short-term result, obtained in an artificial and unconscious way – yes – rather than on the basis of a broad and solid knowledge base?

 There are no easy answers to this question.

 We can speculate that one cause is the capitalism of appearance, where it is enough to look competent and not really be, favoured by the compression of time, which requires the message to be ‘delivered’ as quickly as possible and to generate ‘likes’ and approval. Everything is consumed in a few moments: seconds or a few minutes on Youtube, a little more – but not too much – on radio and television. Skipping is the bete noire of music production: tracks can no longer have too long intros because the listener will skip to the next track. We have to ‘hook’ the listener. Therefore, it is better to use a familiar riff (perhaps because it is ‘freely inspired’ by famous pieces) than an original idea and therefore risk being ignored.

The same distorted logic has spread to teaching. There is no time to teach, because it is necessary to ‘close the quarter’ (the didactic one, not the financial one) and therefore it is necessary to compress the attendance and to lengthen the time: a few lessons that last, each one a long time. How is it possible for a student to metabolise the number of concepts given to him in this way? What learning method could he ever have developed? What conceptual tools will they have learnt to use?

 It is not surprising that such an approach is also practised at work. The availability of pre-digested and homogenised information makes it possible to improvise as an ‘expert’ or ‘competent’ in the most diverse fields. It is enough to provide answers to the interlocutor’s needs, regardless of how or what they have been elaborated, and the actual ability to understand what one is doing. You are expected to make a soufflé blindly following a recipe that has been plagiarised somewhere. As long as all is well, all is well. Then the soufflé deflates and you don’t understand why. The distortion of relying on what has gone before prevents us from solving new problems and, faced with the series 2, 3, 5 and 7, leads us to loudly affirm that even 9 is a prime number.

 In such a context, where knowledge has no role and no value, it is clear that ChatGPT is a godsend for those who work fraudulently. If this is the case, then the merit of the proliferation of AI platforms is to have confronted us with the colossal hypocrisy of having elevated knowledge to the level of a secular deity and then, when the mass is over, returning to the comforting world of superstition, ignorance and fake it until you make it.

 The AI does the rest.

Leave a Reply

Your email address will not be published. Required fields are marked *