Durigon, pensions & AI: the human-machine integration that went wrong

Many underestimate Claudio Durigon, both in terms of his political clout and his expertise. On the former front, he has demonstrated in this Budget Bill that he is capable of standing up to the unyielding Minister for the Economy, Giancarlo Giorgetti. On the latter, just yesterday, Matteo Salvini’s man on pensions showed that he bases his policy proposals on early retirement on a broader vision linking welfare, work and technology. Speaking at the ‘Made for Italy’ forum, Durigon explained that his ‘Quota 64’ – which will be included in the Budget Bill – is a policy for innovation because ‘artificial intelligence accounts for only 8 per cent of our system’; therefore, we must ensure that ‘companies can invest in a young person joining the workforce rather than a 65-year-old’. In short, early retirement – which, under ‘Quota 100’, was supposed to create three jobs for young people instead of one for an older worker (as if that were possible) – now, with ‘Quota 64’, also becomes a lever for technological innovation in businesses: “It improves efficiency for the company and provides an additional right for the worker.” Two birds with one pension.

But since Durigon is not merely a theorist on innovation, he also demonstrates how the legislator can apply technology. Durigon wishes to refute the ‘incredible fake news’ claiming that the government has ‘taken money from pension equalisation to fund the “Quota 64” scheme’, even though the savings from cutting pension indexation (885 million per year) match the fund for early retirement payments (850 million). Durigon not only denies that one measure is used to fund the other; he denies the very existence of the cut to indexation. And he does so thanks to AI. “Just ask ChatGPT something very simple,” says the Under-Secretary for Labour. Type: ‘2020 pension adjustment table’; and at the same time type: ‘upcoming 2027 pension adjustment table’. If you place these two tables side by side, you’ll see that the 2020 table is worse than the 2027 one.” The combination of human and artificial intelligence demonstrates, in a matter of seconds, that in 2020 – “I didn’t choose this at random; it was the Conte-PD government” – pensions increased by less than they will in 2027.

There is a small problem: inflation. In 2019, it was low, so much so that the pension adjustment rate was 0.5 per cent. In 2020, due to the pandemic and lockdowns, there was even deflation, and so the adjustment was zero. This year, however, the realised inflation rate stands at 3.4 per cent and the FOI price index, used to index pensions, is at 4 per cent. It follows, therefore, that a reduced pension adjustment coupled with higher inflation has a greater impact: pensioners grasp this straight away, without needing to consult ChatGPT. The Durigon case demonstrates once again that AI does not solve problems, but rather amplifies the strengths and weaknesses of natural intelligence.