Will AI Replace Your Job?

The big question that those spectacular developments raise is, will AI make us obsolete or will AI replace most of our jobs?

In the past two years, we experienced very spectacular improvement in what AI can do, and in particular, what generative AI can do. Can think of tools like the GPT copilot or mijo. Those tools are now able to do tasks that we saw were completely protected from automation, tasks that are very creative, tasks that require some skills, including analytical skills.

So the big question that those spectacular developments raise is, will AI make us obsolete or will AI replace most of our jobs? So how do we answer this question? The way we answer this question is we look at separate occupations, and we define occupation as a collection of tasks. An occupation is made of several things that we do every day.

Some of those tasks are highly creative, some are very repetitive. Some are very important for us and some are very important for our employer. The way AI is going to compete with humans is not at the occupation level, it's at the task level. So each separate task is more or less likely to be exposed by artificial intelligence based on what exactly this task is.

What we did is that we collected information on 17,000 tasks, and for each of those tasks, we look at the description and we assign the score of how likely an AI will be or is actually currently capable of doing the same thing as a human or even better than what a human would do. We did that for all the occupations in the labor markets, and what we found is that at the end of the day, there are very few occupations for which all of the tasks or a very large majority of tasks are actually exposed by AI.

If you think, for example, of a photographer, a photographer is someone who is going to be directly impacted by tools like me, journey or generative AI more generally, but a photographer is not only producing, you know, editing photography, or generating the background of the photography. A photographer is also trying to capture a moment, discussing with his model, setting a world setup for his photography. All those things are tasks that are very important for the occupation as a world photographer, as a world. And it's not likely that AI is going to be efficient at automating all of those tasks.

So at the end of the day, when you test this task-based approach, you realize that there are not so many occupations that are going to be directly threatened by AI. We found that 20% of those occupations are actually exposed, but it's likely that not all of them are going to be replaced because it may be too costly to do so.

So at the end of the day, we think that the aggregated effect of AI on the labor market is not going to be dramatic in terms of how many jobs are going to be destroyed, but the impacts are going to be very important in terms of what will be the nature of our jobs and how this will change the way we work and the way we organize our different tasks.

At the end of the day, it's also likely that AI is going to create new opportunities, increase the size of the economy as a whole, and so generate new jobs and overall have a positive effect.