Home / Blog / Technology Predictions 2022: Trustworthy Artificial Intelligence

Technology Predictions 2022: Trustworthy Artificial Intelligence

Schematic representation of the various components involved in the delivery of a trustworthy artificial intelligence. Image credit: Policy and investment recommendations for Trustworthy AI. EU

Artificial Intelligence has been expanding rapidly entering in more and more sector. As we marvel at the progress concerns are also voiced on trustworthiness of AI in general and its application in specific, sensitive, domain. The issue is bound to increase as AI becomes more performant (read autonomous) and widespread.

We had a first, clear signal of an AI that has exceeded our “control” in the game of Go, when back in 2016 AlphaGo move 37th baffled (expert) observers who noticed that such a move (a brilliant one, it turned out) would not have been made by a human. The first reactions pointed to a (although limited) supremacy of AI. However, the real issue was not that AlphaGo managed to come up with a baffling move, rather that we do not know how it came to cast that move! In other words: we have lost transparency. Machines (AI) can behave in ways that we did not “programmed” nor that we understand.

As AI becomes more and more powerful we can expect to run more and more into cases where decisions/actions taken by a machine are not foreseeable nor we can understand why such decision was taken (we can only gauge its results…).

You can trust something you understand, possibly you can trust something that has got a very good record (that’s the case of our trust in a doctor, we may not understand her decision because we lack her knowledge but we trust her because of her track record). In case of AI we would like to have at least a few human “experts” that can assure us everything is fine.

There are actually plenty of situations where we trust something we don’t understand, think about flying… The pilot trust the ILS system for a safe landing in zero visibility. The pilot himself may have very little technical expertise in the hardware and software at the core of the ILS system, the trust is based on repeated experience and testing (performed by trusted parties).

There is a strong demand rising for ways to have trustworthy AI. It is just a tad more complicated than trusting an ILS:

Trustworthy AI has been identified as one of the leading, emerging, areas for 2022 and the following years by the group of IEEE Computer Society experts who prepared the 2022 Technology Predictions.

It is noted that there are a number of worldwide initiatives towards a converging set of ethics principles (e.g., secure, private, inclusive, human, responsible, robust) and studies looking at the whole system supporting AI, from hardware to software. At the same time it is noted with concern that the main pressure towards AI evolution is coming from emphases on performance rather than trust.

Nevertheless, it is foreseen that the raising concern of AI provides business incentives, and opportunities, to tackle the issue from the very start, at the design stage. This shall include the goals of explainability, replicability, robustness, privacy protection and fairness. An AI system that can be proven by independent parties to stick to those principles should have a higher market acceptability, hence it should steer more and more companies to adhere to those principle. The forecast is that the time is mature, both from tech standpoint and from demand side, to focus on trustworthy AI in the next coming years.

As the EU noticed (see the graphic) trustworthy AI requires many tiles to fall in place, hence it requires a much broader consideration to become real.

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the New Initiative Committee and co-chairs the Digital Reality Initiative. He is a member of the IEEE in 2050 Ad Hoc Committee. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.