Copilot is not auto-pilot… it keeps humans at the centre: Frank Shaw of Microsoft

We started through the lens of how people really want to interact with technology in a new way. We specifically thought of this concept of a copilot. The first one, of course, was GitHub copilot for developers, which was a way they could have a programming partner to help them write code more efficiently and more effectively and correct things as they went along. And we love that concept, because it’s not an auto pilot. Again, it really does keep the human at the centre of it. If you have a co-pilot, it doesn’t mean you have to stop paying attention. And so we started off with being as a copilot for the web, and then with Microsoft 365, copilots in teams and in word and in PowerPoint and Excel, and then in security.

So our intent is to have copilots that help people with whatever they are doing across all the products that we have, whether it’s Windows, or Microsoft 365, or the mobile devices or search, just something that helps you get what you want done, what you want done more easily and hopefully with a little bit more fun.

Microsoft has had an AI layer with Cortana for a few years now. But do you think AI is suddenly peaking because there is a sweet spot now with technology and products like ChatGPT and so many companies working on so many different things?

That’s right. For most people it feels like it happened all at once. But really, it’s a result of 10 plus years of work and research. And yes, Microsoft has been offering AI products for years. We talked about it as every day AI, whether it’s auto correct in Word or designer in PowerPoint to help make your deck prettier.

But these new large language models are so much more powerful that for the first time they opened up a new interface which jacked up the ability to just talk to your computer or your phone in the way you would ask somebody questions and get answers that we couldn’t do before.

We have all gotten used to modifying how we work based on the limitations of the computer and now we have this opportunity to have computers respond to us in the way that we are most good at. Humans are great at asking questions. Computers have not been great at answering those questions. And I think we have all been in a situation where we are trying to search something and you just type in like a bunch of words and then you are just hoping that you are gonna get something back. Now you can ask very specific questions and then if you don’t get the right answer, you can have a follow-up question. You can do the same thing with a Word document. You can ask questions of a Word document or a website in a way that just feels so much more natural.

So it’s a much wider Canvas in which like people are not restricted by their ability, right?

It will help everybody in different ways. Somebody’s an engineer working on product documentation, who previously might have really struggled with writing the documentation because that’s not what they love to do. Now they have a writing assistant that can help them do that much more cleanly. And as a result, everybody’s going to be better for it. We will see some unexpected uses too. Because when you give people new tools they do wildly creative, unexpected, wonderful things.

Are you already seeing those kinds of new use cases emerge?

We are starting to see some of them and we don’t think we’re the only people to have the answers and if need be. We think of ourselves as a tools company. Microsoft is a tools company. Microsoft is a platform company. When you give people tools you want them to build on our platform, we are not the only ones that have good ideas.

But there are some specific things that we’re seeing that are really interesting. The world today has a security problem, and there are bad actors working harder and harder to get access to people’s information and company information. And there’s not enough security people right now, and there are hard to train. So we have a security copilot that helps level the playing field and give security researchers the ability to move at machine speed in a way they couldn’t before. So that’s an example of doing something completely new and evolving.

I keep hearing the phrase Super Charge and you use the phrase superpower… do you also need some guardrails here, especially when you are looking at new technologies like this?

It was seven years ago that we published for the first time how we wanted to think about responsible AI. And we have continued to evolve and published two versions of our responsible AI standard. We understand that you can’t just slap responsibility on something after it’s done. You actually have to build it responsibly. And we are committed to doing that and we are committed to learning and sharing that learning with the world as well.

And then as we build these products, you have to test them, you know, through the lens of what will people do with them. Brad Smith, our president, wrote a book some years back called Tools and Weapons where he reflected on the role of technology is both a tool and a weapon. And so when you are creating a new tool, you have to think about how it might be used. People are ingenious and creative, also they can be ingenious and bad. And so when we see that happening, we have to change our products to make sure that we’re staying ahead.

How do you judge the productivity of a worker in the times of AI?

We are certainly looking at it through the lens of AI as well. GitHub is probably the most studied, where developers talk about how they are 30% more productive when they are using GitHub copilot and there’s data that supports that. And we are looking at research now that looks at how does a copilot help you at work.

But I think you have to be careful that it’s not just can you do something more rapidly, it is can you do something you couldn’t do before. Does this give you the ability to automate some things that was drudgery before so that you don’t have to do it so that it doesn’t feel like you are just on this endless treadmill going faster and faster. We actually want to get out of the way things that are in between you and your job. I don’t go to work to go to meetings, I have meetings because I want to accomplish something. So how can I make that meeting more efficient, so that I can get to what it is I want to do. People don’t come to work because they want to make a PowerPoint presentation. They want to persuade something. How do I take the drudgery out of that?

So there’s all sorts of things that we can do that will make people more productive without making them feel more overworked. If at the end of the day, people feel like all this is just making me work harder, we will have failed.

Listen to the full podcast coming Monday