How AI can boost productivity

Last week, I had my global leadership in town for our annual planning summit — about 100 leaders from around the world. We actually had a little competition where groups used generative AI for a hypothetical board meeting presentation and gave a pitch on responsible AI use cases. You had 15 minutes to prepare, then ready, set, go.

Each table examined the personalities on the relevant boards, how they might think about AI responsibility, then matched that to the revenue and where there were business opportunities. Then they matched that to our products and capabilities. Finally, they told us a bit about the work we’ve done with that company so far and how that might come together, and created a presentation from it. It’s a great example of a theme around “no more first drafts”. Which is incredibly empowering for anyone in the knowledge sector.

Given AI crosses so many functions across an organisation, what guidance do you have for boards on where accountability for AI should sit within an organisation?

It’s very important to have a governance team established inside an organisation that is grounded in a few things. Not just AI-first, but grounded in data and data governance. Because one of the things you start discovering with these AI capabilities is that they find the chinks in your armour faster than human beings can. Kim Krogh Andersen brought it up in his scenario.

We also saw it in our support scenario when we first started testing that AI persona in our support environment. We found the Copilot was providing some bizarre answers. Why is he giving us the wrong answer for a seemingly obvious problem?

The cool thing with these copilots is that unlike your kids, they will actually show you their homework. Using AI and iterative refinement around data governance and privacy concerns, testing that and hardening it, that is a board imperative.

The one question for boards to ask is, what is your data governance strategy? Closely followed by, what’s your cyber strategy? Because if you don’t have a solid cyber foundation in place — from endpoint protection, through the identity layer, and event and incident management, through exposing and digitisation and AI, business processes can be super-risky.

Microsoft has set up an organisation-wide responsible AI council that’s feeding into the CEO — and then has a board subcommittee particularly focused on AI. What is the board subcommittee’s specific role?

Like any board function, it’s nose-in, hands-out. Making sure you’re asking all the right questions and doing the appropriate coaching. Our work has been really grounded in transparency because it’s a two-way street. I always say, boards do a reasonably good job of solving problems they know about, but a terrible job at working on ones they don’t. It’s sort of an obvious thing.

At Microsoft, our responsibility to the board is to have transparency on all the responsible AI work. For example, we will go through it before any product launch, any go-to-market effort or any actual co-innovation work we do with a customer. Even the work we would do with Kim at Telstra — beforehand, we would say, “OK, let’s release that into the wild”.

The independent, responsible AI governance team evaluates that use case. Then they report against that and we submit that (report) to the board and audit committee. So that level of transparency means you might delay a product launch. It may mean you delay revenue, but if you’re serious about governing AI — and making sure that it is used for good — then that is what you have to sign up for.

How can boards deliver productivity and employee experience gains?

At Microsoft, we measure just about everything. We have a product called Viva that actually helps in these measurements. Viva is a product that existed before all these AI capabilities we launched in market. It was always geared around both employee productivity and satisfaction — measuring yield, efficiency, standard deviation and performance by role in an organisation. Also, measurement of KPIs around wellbeing.

This was a big focus for the product through the COVID-19 era, because it was one thing to measure productivity, which actually spiked during COVID-19 — all records for productivity were shattered — but employee wellbeing went through the floor.

So, how do you measure that and ensure you maintain that delicate balance? Viva helps you do that through your own KPIs and systems of record.

We’ve turned that on the AI utilisation and measured the same things. First of all, we assess what is the daily or monthly active usage of these tools.

Interestingly enough, we see probably the highest active usage inside teams — bringing people together more inclusively than ever before. When somebody showed up five minutes late to a meeting or missed a meeting altogether, we could summarise the meeting for them, bring them along on the journey and help them continue to work with the team on the project.

Things like that have a high degree of utilisation, like the daily savings in your inbox. Triage this mess for me. What’s junk? What do I need to respond to fastest? Did the boss call? Those kinds of things.

Then we have the creation assets. We measure all of those things and the output on productivity, but then we also measure the satisfaction.

Are people, because of that, able to have more free space on their agenda, to have more moments where they can actually freely think? Rather than move from meeting to meeting to meeting and just increase the grind? Even recommendations to employees and managers, like, “Hey, your folks are working 70 hours a week, back it off!” 

This article first appeared under the headline 'Next-Gen AI’ in the May 2024 issue of Company Director magazine.