Which AI risks matter the most?

The MIT AI risk repository asked 272 international experts which AI risks matter most, how severe they could become, and who is responsible for addressing them. 

Experts included AI risk and governance practitioners from financial services and other large corporates, AI safety experts at AI Safety Institutes, academics, and AI policymakers from national governments.

The key findings included:

1. 18 of 24 risks have more than a 10% chance of causing catastrophic outcomes (e.g. more than 1 million human deaths, more than USD $100 billion in damage, or civilization-scale intangible harms such as the collapse of democratic norms or privacy) by 2030 under a business as usual.
2. Mitigations would help but some severe risks remained above 10% and all 24 risks had >5% likelihood of catastrophic harm.
3. Those most vulnerable to AI risks are not those most responsible for addressing them.
4. Information, finance, and national security are the most vulnerable sectors.
5. AI risks can be mitigated, noting that: Significant risks require substantial mitigations; AI needs rules and systems to enforce them; Some risks may need international coordination and structural reform.

The report is also useful as it allows readers to interact with data, visualising it by sector, risk, and actor.

The report comes hot on the heels of a report on how UK public companies report on AI adoption, risks and harms. The reports appear complementary; for example, there is an increasing trend in public companies reporting cyber security concerns related to AI (and the underlying data allows you to see which companies are reporting that) and the MIT report identifies which sectors are affected by those risks, with healthcare, financial services, and national security facing ‘extreme’ vulnerability.

If you would like to discuss how current or future regulations impact what you do with AI, please contact Tom Whittaker, Brian Wong, Lucy Pegler, Martin Cook, Liz Griffiths or any other member in our Technology team.  For the latest on AI law and regulation, see our blog and newsletter.