Lawyers, computer engineers… usually, students who turn to these professions make their families happy, reassured to know that they are on the highway of stable and well-paid employment. But 2023 is an earthquake for white collar workers. The rise of generative AIs such as ChatGPT, capable of producing text and intelligently analyzing large amounts of data, promises to disrupt the economy. And if the profiles with few qualifications had been the main ones concerned by the previous technological revolutions, this time it is the Bac+5 who find themselves in the front line.
THE latest McKinsey report on the economic impact of generative AI is eloquent on this subject. While the situation changes little for baccalaureate level profiles (the potential for automating their function increases from 51 to 64%), the increase is very marked for diplomas at master’s level, doctorate and beyond (it goes from 28% to 57%). “Last November, the main exams of American universities still presented some challenges for GPT-3.5. The most recent version, GPT-4, immediately ranks among the top 25% of candidates”, points out Eric Hazan, associate director senior at McKinsey.
Bac+5 and the ChatGPT revolution
The cradle of generative AIs, the tech sector will also be among the first to be upset by their capabilities. Because these AIs do not just produce texts or images, some like Github’s Copilot are also very good at “generating” computer code. “Copilot analyzes the code that a developer is typing in order to guess the lines that he will need to enter later and to offer them pre-entered. The AI therefore saves coders a lot of time “, explained to us last October Thomas Dohmke, CEO of GitHub. Proof of the tool’s effectiveness: at that time, already 40% of the code written by people equipped with Copilot was written by AI and no longer by humans.
The banking sector could benefit even more, points out the McKinsey study. According to the firm, it is even he who could derive the most benefit from generative AI, in particular by personalizing his advice and his exchanges with clients. “Morgan Stanley is building, for example, a smart assistant based on GPT-4 to help tens of thousands of wealth managers quickly find and synthesize answers from a vast internal database,” the report says. .
In all, 70% of businesses should see more than 50% of their tasks automated by 2045, according to the report’s median scenario. Figures that are enough to make white-collar workers dizzy. Will they be “uberized”? The horizon is actually less threatening than it seems.
The fact that generative AIs have become masters of human language strikes a very sensitive chord with us. The content and information they produce, however, desperately needs to be controlled by flesh-and-blood humans. Lawyer Steven Schwartz had a bitter reminder. This American had instructed ChatGPT to search for legal texts likely to help him support his case. However, the AI sent him opinions and references made up from scratch that Steven Schwartz did not take care to check before using them. A false step which earned him to be summoned before a judge on June 8, reports the New York Times.
Of course, companies could be tempted to take advantage of all these hours of work delegated to the machine to reduce their workforce. Some certainly will. As the competition increases its productivity with these tools, some will however choose to devote this additional time to more sophisticated achievements. “With the advancements that AIs had already made in recent years, many expected the number of legal assistant positions to drop sharply in the United States, this has not been the case at all”, confirms Eric Hazan Firms are now actively seeking paralegal knowing how to use AI to quickly find the relevant case law. And instead of asking for three examples to back up their case, lawyers now ask for dozens, with the AI creating richer documentation in the same amount of time.
“In countries like France, given the aging of the active population and the new jobs that will be created by these technologies, there is no risk of creating significant unemployment. We will be a net creator of jobs. However, the The positions that will be created will not require the same skills as current jobs, “says the senior associate director at McKinsey. The era of easy summaries and presentations that brew wind is probably living its last hours. Phew. Generative AIs us however, bring us back to the disturbing question: what more can I do than the machine?