About the Future of Work & Innovation Cycles
New study sheds light on what kinds of workers are losing jobs to AI
https://cbsnews.com/news/ai-artificial-intelligence-jobs-workers/ CBS News
Artificial intelligence is replacing entry-level workers whose jobs can be performed by Generative AI tools like ChatGPT, a rigorous new study finds.
My take here:
Exceptional study here on the Future Of Work and in particular on the impact of LLMs on the replacement of humans by AI.
Excellent study given that it is very thorough and detailed and that generative AI is very recent, having only appeared in 2022 with ChatGPT. As I already pointed out at the beginning of this post, this study is very serious when it comes to the future of work. However… I disagree with part of the analysis of the designers of this study here.
As a specialist in innovation cycles myself (read or reread, if you are not too familiar with this field, for example, the economist Schumpeter on this subject : https://ecoholics.in/schumpeter-theory-theory-of-economic-development-through-innovation/ ), yes, of course, previous innovation cycles have all been “destructive” of jobs and then “creative” of jobs.
The authors here of this study (like some other analysts in the sector) believe that this new wave of AI innovation will inevitably follow the same path of job destruction/creation. I disagree. Why?
I will give you a figure (which is an average, and therefore not really significant in microeconomic terms, but very significant at the macro level, and therefore in terms of overall labor market trends).
This figure is very easy to analyze: AI will/can replace six employees, so one AI is enough to do the work of six humans (on average across all jobs and sectors, with significant disparities depending on the sector, of course). But the five humans who have lost their jobs, even if they train in AI, will not find another job, simply because those other jobs will already be filled by AI. The difference here, compared to previous cycles of innovation, is that this new economic cycle has not at all the same “technical” nature as the previous ones, which consisted of pure automation coupled with a work organization designed to streamline and maximize production costs. Here, with AI (and now LLMs), a business leader will never replace the five employees they laid off, since AI, for example, will not get sick (or at least not as often as a human!), will not ask for raises, and will not go on strike.
Furthermore, in previous innovation cycles, the sectors that created major innovations (Taylorism, Fordism, and Toyotism, for example) were huge providers of jobs. Today, with AI companies, this is no longer the case. The companies behind this new cycle of innovation generate huge stock market valuations, but very few jobs. This new innovative cycle of Digital Transformation is clearly different in nature from the previous cycles theorized by Schumpeter. And its effects on employment will also be different, totally different. Worrying? Yes, obviously.
