'The AI Act is not a brake but an opportunity, regulations are also needed for defence'
Mariarosaria Taddeo, Professor of Digital Ethics and Defence Technology at Oxford, says: 'Technology at the cost of rights is not desirable'.
The controversy over the AI Act is 'induced' because it is not so much about regulation per se as it is about bureaucracy, which certainly needs to be streamlined. However, regulation is important, because a technology, such as artificial intelligence, that develops 'at the cost of our rights is probably not desirable'. This is stated by Mariarosaria Taddeo, author of 'War Code. Ethics of Artificial Intelligence in Defence', Professor of Digital Ethics and Technology for Defence at the Oxford Internet Institute at Oxford University and Director of the UK Digital Ethics and Technology for Defence Research Group. In the interview with the GENerAzione AI Observatory, in collaboration with Accenture, Taddeo emphasises that it is important to put in place not only regulation that restricts the use of AI, but also incentives for investment. And on the subject of the defence sector, which remains excluded from the AI Act, she calls for rules to protect human rights and democratic values.
Professor, there is much debate on the AI Act and, more generally, on the European over-regulation of artificial intelligence, which some claim is holding back development and investment. Do you agree?
I think the controversy over the AI Act is partly induced, and not so much about regulation as about the excessive bureaucracy that comes with it. I think regulation is important, not only because a technology or economic push that comes at the expense of our rights is undesirable, but also because regulation creates security in the market. And this attracts both investors and technological innovation. The point, if anything, is that European regulation is biased: we regulate to limit risk, but we have so far seen very little strategic approach to incentivising European innovation, technology and investment in the sector. That aspect of regulation should be strengthened.
Italian small and medium-sized enterprises - which make up the vast majority of the business fabric - complain about the bureaucratic burden...
The bureaucratic aspect needs to be streamlined, not least because European regulations are supplemented by national regulations, creating a complex ecosystem that, frankly, one could do without. But streamlining bureaucracy is not the same thing as streamlining regulation. This should be clear. I am not a jurist, but I think processes can be streamlined while maintaining the fundamentals. From here to say that the AI Act is 'against innovation', however, there is a long way to go. In fact, I believe the opposite: innovation is only such when it enables us to live in better societies and lead better lives. If it does not help us achieve these goals, it is not innovation: it is an instrument of regression. The AI Act applies to AI systems that pose serious risks to our societies. The Commission expects between 5% and 15% of AI products to be classifiable as high risk. For the companies that produce them there are stringent obligations, for the others the implications are minimal if any.
