Artificial intelligence, robotization, the knowledge economy, the big resignation: these are forgotten in the pensions debate.

Atlantico: To what extent do the COR scenarios regarding the balance of the pension system, be it artificial intelligence, robotization, the knowledge economy, or the great resignation, ignore the fundamental changes facing French society?

Lawrence Alexander: Scenarios produced by COR, by definition, will not come true. They do not incorporate the radical technological changes, the economic effects of artificial intelligence and robotics. They do not include the structural changes in the economy associated with these developments. These are interesting scenarios for short-term macroeconomic management, but they are not predictive in the medium term.

Pierre Bentata: Although COR reports are well prepared, they are often misinterpreted or misunderstood. But the whole difficulty of any forecast is that they are limited by the level of knowledge at a moment, which forces a static analysis. Thus, there is a global rationale that we cannot incorporate major structural changes for which we do not know the effect of COR precisely. Technologies will affect our operations and organization in two ways: by changing the way we work, it will affect employment, productivity, competition between workers and, ultimately, wages. But the second effect is the increase in life expectancy that technology will create, which inevitably affects pensions. The heart of new technologies has not really moved: nanotechnology, biotechnology, computer science and cognitive science. It competes with individuals in their ability to perform complex tasks. And if we cannot integrate it into our scenarios as it is, it should be reflected in the pension funding system. Another decade of life expectancy would make our pension system completely unsustainable under all possible COR scenarios. It is no longer a question of 64 or 67. Especially since every major technical development is accompanied by a decline in demographics.

Read also

5 embarrassing questions to figure out how much pension reform will cost us

Can we predict and quantify the consequences of these changes for the French economy?

Lawrence Alexander: No, because it is not possible. When it comes to AI, we never guessed the winter and spring of AI. We were very optimistic after the film 2001 a space odyssey we believed in strong artificial intelligence for horizon 2000. And nothing happened. Conversely, when deep learning applied to image analysis developed, no one believed it anymore, but the increase in the size of datasets allowed unexpected advances in 2011. Similarly, we haven’t seen a boom in Large Language Models, LLMs. Chat GPT 2 was bad, the 3 we have now is great. He makes biographical mistakes, is not very good at math, invents elements, etc. But it is not surprising at all. But it is impossible to predict the impact of technological innovations on pension systems, because we are unable to predict when technological upheavals will occur. For example, there’s no telling if ChatGPT will grow much.

After AI, robotization is anecdotal. It’s slow going because it requires a lot of rare metals and you can’t rework robots overnight whereas you can for AIs. The inertia of robot developments is huge. Therefore, we can doubt the mass robotization in the coming years. On the other hand, partial automation in the third sector thanks to AI is more plausible. Professions will be shocked, and journalism will not be left out. ChatGPT can now write average articles in record time. Buzzfeed has already announced that it will use this AI to help create content.

Read also

Retirement: What government and unions can (or will) agree to avoid impasse

Pierre Bentata: It’s hard to quantify, but we know the trends. Our analysis of previous innovations shows us unfavorable scenarios. We know that in the medium term, technical development creates jobs and creates more than destroys them. They are changing the structure of jobs to make them less difficult, more skilled, and less paid. And those who don’t embrace these developments find themselves in service jobs, but paying less than before. On the other hand, there is always a period of friction. When machines suddenly replace non-substitutable occupations, it makes some workers “obsolete”. That’s ten years of real losses. Studies such as The Future of Employment by Carl Benedikt Frey and Michael Osborne or The 4^c The Industrial Revolution and the Future of Jobs by Nick van Dam analyzed these events. The number of jobs always increases and the population benefits from general enrichment, but this does not happen immediately. And when that happens, it creates real problems.

As far as I know, we cannot estimate the impact of the knowledge economy or major technical developments. On the other hand, what is certain is that if the decision-makers were opportunistic, they would have an additional argument for the capitalization share, because there is a risk of negatively affecting the distribution, but the enrichment of advanced companies. . They could use a wave.

Read also

French schizophrenia: obsession with retirement age, widespread indifference to elder mistreatment

As for the big resignation, we don’t feel it in the data yet. We do not see the formation of an alternative system. The main challenge is meaning at work, and technologies will only enhance it. And that will be the case from this training. Chat GPT can already do an impressive job, and many skilled jobs will either be replaced or require more skills. The feeling of being in competition with the machine reinforces the loss of meaning. The problem with new technologies is that you never know what impact they will have. TikTok is becoming a rival to Google that no one even imagined. To avoid machine slavery, decision makers must be proactive.

Michael Ruimy: The fourth industrial revolution involving the introduction of new technologies requires a workforce with broad in-depth knowledge/skills that can easily grasp new jobs. When discussing the impact of automation and the dazzling advances in robotics and artificial intelligence… on employment, existing research shows that it promotes inequalities and already has enormous implications for learning: Many children entering primary school now have skills that do not exist today. work on jobs. This raises concerns about a future without employment (actors lacking the necessary human capital will see their labor market integration weaken in the long run).

Already, the global stock of industrial robots has doubled over the past decade, with France having about 150 robots per 10,000 workers, Germany just under 350, and Singapore nearly 830. Robot technology must continue to evolve: the number of robots will increase. Four times in the world by 2030! When it comes to artificial intelligence (AI), its main expected impact is not the destruction of jobs, but the reorganization of work that the spread of technology should bring. According to the OECD, about a quarter of jobs will be affected by this redistribution of duties. A decade ago, a study examining the expected effects of increased computerization of occupations in the United States on labor market outcomes concluded that less than 50% of total employment in the United States fell into the high-risk category, particularly those in the transportation and logistics sector, as well as the majority of office workers ( jobs that can be automated fairly quickly or even within the next few decades). The risk of witnessing unstable career profiles should lead to a reflection of social protection adapted to the new model promoted by the digital sector.

So in the big debate about balancing pension systems, should robots be taxed appropriately to save our pensions? In other words, if robots are taking our jobs, should salary contributions be taken from machines? The idea that animated the 2017 presidential campaign. If robots can replace humans in the work context, there could be a gradual decline in work-based social payments, which could affect the financing of pensions. But in order for robots to contribute to our pensions, this requires social security contributions to be made, but above all they must be paid a “salary” against which these contributions will be applied.

Ethical and legal questions may arise if we consider the equivalence between the cost of using a robot and the cost of traditional labor. There is something unsettling about comparing a robot to a human. In addition, in order to assign even a nominal salary to a robot, it is necessary to give it the status of a legal entity. Finally, the current intergenerational system is based on solidarity: pensions for the elderly are paid for by working people. What will the system be like when a person is replaced by a machine? Robots are “scrapped” and do not benefit from retirement. But the fact is that if the productivity achieved by robots advances their use, additional turnover will be found in tax receipts, and there will always be time to transfer some of this to pension funds.

How can we approach the topic and these changes in our thinking about pensions?

Lawrence Alexander: Even if it is difficult, we need to integrate more technological elements into COR scenarios…

But technological reflection should not be limited to determining the future deficit of pensions, but at the same time it is necessary to prepare all the necessary changes. Company, school and hospital must prepare!

Pierre Bentata: We do what we do when we encounter complex systems. When the system itself evolves through endogenous mutations that depend on the evolution of the system, it is impossible to predict exactly what will happen, but we can think about the main trends and risks: increasing life expectancy, changing certain jobs, etc. We can’t foresee everything, but we can think about it to be more resilient and allow the system to continue.