News analysis
Boards of AI companies: what should they look like?
Boards of AI companies have a lot to worry about as their performance goes further under the microscope.
Ultimately, there’s no avoiding the extra scrutiny because AI is one of the significant changes of our generation – stakeholders want to ensure the companies providing it are run correctly.
The trouble is that each group of stakeholders has its own definition of “correctly”; boards need to find the balance.
What’s the latest?
Since the chaotic dismissal and re-hiring of OpenAI CEO Sam Altman, more attention has fallen on how AI firms are governed.
At the heart of any company like OpenAI is its board of directors, which is supposed to govern in the company’s best interests. If they don’t, they are in breach of their fiduciary duty.
If that’s not enough pressure, the spotlight is now firmly on AI companies and will be for the foreseeable future.
Is that a bad thing?
Only to the unprepared. Remember: this is AI we’re talking about. Its impact on society is so enormous that it was only natural for major attention to fall on the industry and the boards working within it.
With all that focus, it’s natural that the pressure and opinions will increase.
Take recent comments from Yoshua Bengio, a co-winner of the 2018 Turing Award (often considered a “Nobel Prize of computing”).
He recently stressed the importance of having members of the public on the boards of AI companies to provide democratic governance that would adequately represent society and protect it from runaway developments.
“We need an inclusive set of people on the board of these organisations, who can have visibility on what is going on and can act in a way different from the regulators,” he stressed.
Is that kind of talk realistic?
Plonking members of the public amongst a board of executives and expert non-execs would undoubtedly be disruptive.
Depending on the company, it could create practical problems and come across to other directors as a nuisance before they begin to notice tangible benefits.
But…
The underlying principle is more realistic because it reflects one of the core issues that boards of AI companies should absolutely be concerned about in the coming years.
What’s the issue?
They need to recognise just how important AI is going forward.
The industry will command global scrutiny for years – maybe more – as it leaves its inevitable and lasting impact on human life. Such an impact will be monitored closely by various stakeholders with differing expectations and levels of trust.
The leaders of these companies will be closely monitored, too.
That’s why AI boards must take steps now to be as open, transparent and representative as possible. These things are essential in any governance scenario, but getting it wrong in an industry as crucial as AI is a major mistake.
We’ve seen it in the Sam Altman case: a cagey board, with little stakeholder representation, giving severely mixed messages, almost led OpenAI into disaster. No company can afford this chaos, especially not when the world’s media and AI critics will descend like hawks once they get the idea that something’s wrong.
Okay – Bengio’s specific idea of director public representation might not land well with every single board involved with AI. Still, it should give them good food for thought: how will we ensure our board has the voice it needs to function correctly?
Ultimately
These days, many boards already strive for more openness and representation, so the need for it in AI is nothing new – but it is urgent.
The world is fascinated by this one industry and anxious about how it will shape people’s lives. Governance mistakes in an arena like this are not ideal.