Seven lessons from the history of innovation to better understand the potential of controversial AI tool ChatGPT
Unless you’ve been living on Mars for the past few weeks, you can’t avoid ChatGPT, an AI tool that answers all your questions: article summaries, economic crisis summaries, poetry writing, and more. Like any new technology, it is presented by some as revolutionary, and by others as useless, useless, even dangerous. Although it takes time for the light to shine, we can avoid some pitfalls and above all entrenched positions by relying on the history of innovation, which offers at least seven lessons for a more nuanced approach to the debate.
The history of innovation is complex. It’s full of technologies that promise a great future but deliver nothing. Others took years to succeed: the first automobile was born in 1765, but we didn’t get one until the late 1880s. Others, after all, were born into neglect, completely underestimated in the beginning. The Wright brothers’ first flight in 1903, a historic event, got three lines in the local newspaper, and it took a long time for aviation to be seen as anything more than a whim of the wealthy. The greatest minds have misjudged the potential and impact of new technology, and the author of these lines is not about to add his name to that list. However, there are seven historical lessons to better illustrate the potential of ChatGPT.
1) The success of a new technology is rarely the result of its technical performance alone. Diffusion is a truly social process: a social body accepts a technology, but its use generally leads to its modification. So it’s a complicated process. A very good technology may therefore be rejected due to social criteria associated with dominant mental models. In France, for example, GMOs are rejected because there is a mental model of “GMO = danger” that successfully opposes the “GMO = great solution” model. Therefore, it is impossible to predict the economic, social and political consequences of the new technology. What happens to technology is the product of both a technical (its invention and improvement) and a social (its adoption, rejection, adaptation for sometimes unexpected uses) process. When we think about the potential of ChatGPT, we cannot limit ourselves to discussing its technical specifications. For example, we know that some of his answers will necessarily contradict the beliefs of certain groups and this may lead to hostile reactions. ChatGPT will therefore be “moderated” (ie censored), which will lead to other hostile reactions.
2) Any technology, like any tool, has its limits; should not be judged by them. Neither is universal. Thus, the limits of ChatGPT were noted very quickly: lack of references, questionable positions, sometimes strange results, lack of creativity, weaknesses in certain tasks, etc. Understanding these boundaries is key to determining where technology will and will not be appropriate. It is a mistake to point to these limitations in order to dismiss it altogether. Drawing software can’t make you an artist, so it’s not useful. What is needed is not to reject technology for its limitations, but to focus on its potential, to understand what it allows us to do again.
3) There is a tendency to judge new technology with existing technology. However, new technology introduces new performance criteria and must be judged against these criteria. 3D printers don’t offer the quality of traditional factory production, which is why they’re compared negatively, but that’s not what they’re trying to do. They provide very useful flexibility and customization in areas where the quality they offer is sufficient. Therefore, they are suitable for some uses and not for others. The fact that a new technology underperforms existing technology on some criteria and outperforms existing technology on others thus explains why it rarely completely replaces the latter. We continue to use traditional stoves in addition to microwaves, propeller planes, and not just jet planes, paper planners, and not just electronics, etc. Sometimes it won’t meet historical criteria (a cell phone can run out of battery or lose network unlike a landline), and sometimes it surpasses current technology by all criteria, as was the case with digital photography, for example. 2000s, in which case the transition is complete and the old technology disappears.
4) New technology improves its performance over time. Therefore, in order to evaluate it, we must look not at where it started, but where it can go, which is obviously difficult, if not impossible. The online machine translation service Google Translate was launched in 2006. What a magic to provide English translated text in seconds! Of course, this required a significant rework. Let’s say GT did about 50% of the work, but the time savings were enormous. Then the quality gradually improved. In 2017, DeepL appeared, showing significant progress. The translated text took a few minutes to finish, we reached about 80%. Today, the quality has become remarkable. There is practically no need to rework the text. If we were to judge machine translation by human translation in 2006, we would strongly reject it.
Acceptance and rejection of new technology
5) New technology is first adopted by non-consumers. The quality of machine translation in 2006 was far below that of a professional translator, given that I had neither the time nor the money, neither of which would work to translate three paragraphs here and there. . Despite its mediocre performance, machine translation was already doing me a huge favor. The alternative for me was GT or nothing. Now, as old as it is, the GT is better than nothing; its performance was therefore sufficient for the non-consumer that I am. This is why the new technology is accepted by non-consumers despite its limited performance. He gives them what they could not have before.
6) Those who use existing technology tend to reject new technology. This is a consequence of the previous point. This is due to the fact that the latter is generally not good enough for them. I started using Internet telephony in 1998. The quality was terrible, but I could call abroad for the price of a local call. In other words, the performance of Internet-telephony was enough for me for its value. On the other hand, it was not possible to use it for a company, the performance was not enough for its requirements. “It didn’t work” for him should be understood as “Its performance is not sufficient for my requirements, so I can’t use it.” Quite logically, companies initially abandoned the Internet phone, so it started with personal use.
7) We tend to emphasize the disadvantages of new technology more than the advantages. Talk to robots, we answer unemployment. Talking about biotechnology and genetics, we answer Frankenstein. Talk to IA, we give you the answer to Skynet and machine domination. This is not new. The appearance of the photo led to fears that artists and radio musicians would disappear. We focus on what will disappear or at risk of disappearing without imagining what can be born: Gutenberg’s printed book caused the disappearance of magnificent illuminated manuscripts, true works of art, copywriting monks deprived of a source of income, but began to read. and writes to everyone. With books, a scholar no longer needed to memorize texts; therefore, he could use the freed capacity of his brain for more creative tasks. It is impossible not to think of a possible similar effect for ChatGPT, which frees us from certain tasks and allows us to do others.
homo ludens
The history of innovation is complex, and predicting the impact of technology on society is difficult, if not impossible. Its success often depends on social factors and public mental models, and its potential should not be judged only by limitations or technical indicators. Its success depends on what we do with it, and therefore the strange dance between its promoters and its users. So, if you want to get an idea about chatGPT, start using it to find out its strengths and weaknesses. Take advantage of the former and compromise on the latter; follow its development as a machine translation. In short, ignore the overconfident morality teachers and pundits and play with them. Have fun!
? This article title was found by ChatGPT
➕To go further on the topic, you can read my previous articles: ▶️Evaluate the Disruptive Potential of New Technology: Value Curves ▶️On MOOCs Creating Pschitt… and on disruptive innovation in general. Technology as cognitive augmentation is an idea noted by Michel Serres. Watch here: ▶️ Conference by Michel Serres: technology enhances our cognitive abilities.
? If you liked this article, don’t hesitate subscribe to be notified of future events by email (subscribe at the top right of the home page).
? This article is available in podcast audio format: Apple Podcast – Google Podcast – Tumult – Deezer – Spotify – Podinstall
