How the AI would respond if it were to face Vannacci on live television

If an artificial intelligence were to face Roberto Vannacci live on television, it would probably begin with a revolutionary move: it would let him speak. It would not interrupt him when he says that Italy must return to being "the home of Italians", nor would it grant him the refuge of the victim of political correctness. It would wait for him to finish his sentence and then utter four devastating words: ‘What exactly does that mean?’ The advantage of AI would not be moral superiority. It would be patience. Vannacci is built to win the symbolic battle: identity versus conformity, courage versus cowardice, the general versus the armchair intellectuals. But a machine has no salon to defend and, above all, does not get distracted. If Vannacci proposes reducing the immigrant population to 4 per cent, the AI would not ask him if he is racist. It would ask him why specifically 4 per cent, how that figure relates to the needs of today’s Italy, and who should leave: undocumented migrants, legal residents, those in work, those born here, or those who have become Italian. Then it would ask: how much would it cost? The moment he uttered the word ‘remigration’, the AI would open the can of worms. Deporting a convicted illegal immigrant is one thing; expelling a legal foreigner or revoking citizenship is another. Which of the three? Under which law?

On the issue of security, the system would not dismiss citizens’ fears – a mistake that often wins Vannacci votes. It would, however, distinguish between fear and data. Which crimes have increased? Where? Building new prisons may help, but how much does it cost and where are the police officers to be found? Complexity is not an excuse for the weak: it is the grown-up name for reality. The same approach would work for Europe. Do you want to discuss the euro? Fine: what would happen to mortgages, savings and energy costs? Do you want more sovereignty? How much does a country gain by facing China, Russia and the United States on its own? Do you want to reduce support for Ukraine? The AI wouldn’t ask you if you love Putin. It would ask you what incentive an aggressive power would have to stop if an invasion resulted in territorial gains and impunity. Then it would turn to the family. Cutting taxes for every child is a legitimate proposal. But how much does it cost? Which spending is being cut? Why not also talk about nurseries, women’s employment, wages and housing? And regarding the employment record book at the age of fourteen, it would ask where apprenticeship ends and where early school leaving begins. At that point, Vannacci would probably try to change the subject, denouncing the betrayals of the right, the hypocrisy of the elites and the hostile press. But the AI would not fall for the diversion. “That wasn’t the question,” it would say. And it is this very phrase that could put him in a difficult position.

Not because all his questions are illegitimate: some do highlight real problems. But his method consists of turning every problem into an identity, every identity into an enemy, and every enemy into a shortcut. An artificial intelligence would force him to move from slogans to the balance sheet, from posturing to consequences. It would not call him a fascist, an extremist or a populist. It would present him with an Excel spreadsheet. And this is perhaps the general’s weak point: not the harshness of his responses, but their vagueness masked by that harshness. Vannacci can win against those who take offence. Against those who ask ‘how?’, ‘how much?’ and ‘with what effects?’, his culture war risks ending before the adverts.