Trusted AI: A lot is happening

by Sandy Pentland

Alex “Sandy” Pentland is a HAI Center Fellow and faculty lead for digital society at Stanford HAI and the Stanford Digital Economy Lab. Reid Hoffman, co-founder of LinkedIn, described Sandy’s new book Shared Wisdom: Cultural Evolution in the Age of AI as a “must-read for anyone who wants AI to strengthen human insight, not override it.”

I’ve recently joined a group of Stanford, Berkeley, and MIT faculty working to flesh out the recent US-China agreement to develop interoperable architecture to ensure that the AI agents beginning to appear in trade and finance are trustworthy. It is not just the US and China who want trustworthy AI. Organizations ranging from the African Union to the international task forces that maintain digital safety standards have joined in as well.

Leaders around the world are beginning to realize that not only must we act to prevent advanced AI from being exploited by criminal or terrorist networks, but also that we need to make sure that the rapid deployment of huge numbers of perfectly legal AI agents for commercial services doesn’t cause runs on banks, price wars over commodities, or the collapse of markets.

Why allow AI agents in trade and finance at all? Because when operating properly they can be orders of magnitude more efficient and far more secure and transparent. The promise of better global systems is why this push to ensure trustworthiness in global trade and finance is being driven by national ministers of the economy and secretaries of the treasury. Their responsibility is to keep trade and finance healthy, and so they are focused on keeping markets stable and trustworthy while leveraging AI agents to make the economy more productive and accountable.

As a consequence of this interest, engineers around the world have started to develop practical solutions to the problem of making our most critical systems safe for AI agents. Unlike the more difficult problems of how different cultures can work together to align AI with human society and ensure that AI promotes individual thriving, trade and finance have centuries of international practice that has generated general agreement about how things are supposed to work. This agreement means that there is a reasonably clear technical path to making sure that the flow of food, medicine, and money can remain dependable as AI agents blend smoothly into the world’s trade networks. The challenge, of course, will be to invent solutions quickly enough to keep the world on an even keel, while at the same time looking to the future to ensure we are building an architecture capable of addressing the more difficult societal problems.

In Shared Wisdom I predicted that governments, companies, and multinational organizations would begin the process of creating trustworthy AI starting with the international systems that underpin our societies: trade, finance, anti-fraud, shipping, defense, payments, and health, with the goal of making these must-have-today systems prepared for AI agents. As predicted, the organizations that manage the world’s networks, such as the IETF (which defines the internet’s architecture), FATF (anti-fraud efforts in every country), and the large payment, finance, trade, and communications corporations, are now urgently working out how our society’s systems can remain stable and trustworthy as AI agents become ubiquitous.

I see real progress within my discussions in the EU’s banking center, the Abu Dhabi Investment Authority Lab, the Indian G20 meeting, Singapore’s SMART program, and Japan’s leadership. Their solutions, while likely not always the most elegant, will hopefully keep the world on an even keel for long enough to figure out how to do the bigger job of aligning AI with human values.

It will be disappointing to some that regulation around the big questions of cultural alignment and the facilitation of individual flourishing is not moving as quickly as work on these more practical problems. I am optimistic, however, that the solutions found for more mundane but critical systems like finance and trade will allow us to do a better job of addressing the bigger problems.