Beware of figures on ChatGPT's impact on employment
"Let's be cautious about incomplete figures regarding the impact of ChatGPT. This morning, I read an article shared by Wladimir D. on ADN (https://www.ladn.eu/tech-a-suivre/vires-par-leur-employeur-a-cause-de-chatgpt/). I noticed numerous biases and deficiencies in this article.
It is stated that 48% of companies that integrated ChatGPT have laid off employees. However, without knowing the percentage of companies that integrated ChatGPT, this indicator remains incomplete, as we have X/100 * 48/100 of the companies that have laid off employees, making it impossible to determine the overall impact of ChatGPT. Furthermore, we do not know the average percentage of employees laid off by these companies, AND MORE IMPORTANTLY, the total number of jobs and the size of the company. Ultimately, we are talking about a few percent.
The article also mentions that OpenAI estimates that 80% of jobs would be impacted, but it doesn't specify the scale or manner of this impact. We can question these figures from OpenAI, as they might be inclined to overestimate the impact of ChatGPT for commercial reasons.
In reality, considering the total mass of employees in all companies, the potential impact (which I dare say is highly unlikely) on the economy is much lower than the highlighted 48% figure, probably just a few percent at best.
Another example, IBM stated that 30% of its administrative jobs could be performed by AI. However, this economy is mainly due to the use of RPA (Robotic Process Automation). There are much greater savings to be made by optimizing business processes, but that's another topic.
We should never underestimate humans' ability to create entropy in political contexts (chaos). Announcements about the impact of AI on jobs have already been prone to errors in the past. In 2013/2014, an Oxford Business School study gained media attention, claiming that 50% of jobs would be impacted by AI. Unfortunately, few journalists had actually read the study (only the abstract), and reality completely contradicted these predictions, as the study was full of biases and limitations.
We must, therefore, remain cautious and critical of those who present these figures without proper perspective or in-depth analysis. It is essential to keep in mind that in this matter, we have:
领英推荐
It is interesting to project ourselves into the future by considering a hypothesis: If everyone starts using such tools, we may risk seeing a standardization of production (media), with content that looks the same from one magazine to another. This raises the question of differentiation as a lever of competitiveness.
It is conceivable to abandon the generated content and read only the "prompts" (instructions). Web scrapers could then generate structured data directly from the prompts, allowing resource savings.
However, excessive automation presents risks, especially in generating errors that could expose companies to legal consequences.
As with any technology, we are likely to go through a phase of adoption, excess, and rebalancing in the use of ChatGPT and other similar tools. It is important to remain vigilant and exercise discernment to avoid potential pitfalls associated with automation.
In conclusion, it is crucial not to take the figures tossed around in the press at face value. We must be especially wary of individuals who highlight them to sell us something. Two essential elements are missing: critical minds that dare to doubt or admit their ignorance, and complete, sourced, and verified figures.
Until we obtain reliable information, I suggest paying homage to Pierre Desproges, the french comic from the '70s and '80s, by doing one simple thing: Chuckle!"
Jerome Fortias https://www.nexai.net
Happy Birthday Jerome ??