Name: 'Hard to Predict' Where AI Will Be in 6 Months
Uploaded: 2024-10-03
Duration: 10 min 52 s
Description: Lili Cheng, Corporate Vice President of Microsoft AI and Research, chats with Fortune about building vs. buying AI systems, AI copilots, and the future of the advancing technology.

Lili Cheng, Corporate Vice President of Microsoft AI and Research, chats with Fortune about building vs. buying AI systems, AI copilots, and the future of the advancing technology.

'Hard to Predict' Where AI Will Be in 6 Months

Conferences

tempaLili Cheng, Corporate Vice President of Microsoft AI and Research, chats with Fortune about building vs. buying AI systems, AI copilots, and the future of the advancing technology.

Lily, I think uh what everyone is excited to hear about, hear about is, is you being on the A I front lines at Microsoft? So I wanna start by talking about the big trends that you're seeing in A I model. So what new capabilities in A I systems are coming soon that all of the people here cio s top tech people should really know about. Well, it was great actually talking before, you know, being up here, I think everybody is just kind of immersed in A I and how quickly things are moving. And so, you know, in a way, it's hard to predict what's gonna happen in the next six months even. I mean, we see a lot of trends that you've probably seen over the last week or two, you know, um more types of conversation and media coming in, more asynchronous conversation and the ability to do more um complex tasks. Um I think so far, we've seen a lot of progress in some of our co pilots at Microsoft, like the github copilot. We've seen developers think today over 50% of all code in github is A I generated which is kind of mind blowing for a product. It was one of the first co pilot and we see the developers probably 90% of developers um find that they can do their repetitive task much more easily, but still maybe 60% or 70% of what a developer does isn't coding. It's really around collaboration and um coordinating work um doing deeper thought. And so I think one of the things that you're gonna see A I looking at is how do we address some of the real challenges that people have around getting work done? Um More of the complex tasks and um we'll continue to see progress. I think in the, you know, more tactical task, but I think you're gonna see more um work flows and workspaces people collaborating and some of these challenges being addressed, what are some of the biggest challenges you have faced and that you're trying to solve for Microsoft. Um It's interesting like in a way even at Microsoft, I think A I cuts across so many different groups. So it's causing people to work together in a really different way. So our whole company, I feel like it just collapsed into, you know, a group of people who have to take on more discipline. So the difference between like what a product manager does and a developer does a machine learning expert, a designer, something working with customers. Um We're just seeing a lot of things merging together and I think that's partly because of the way A I um is so general purpose and it's, it's merging like workflows, things that were vertical before you see things um wanting to combine. So an example is we have a product that does sales, we have a product that does service. And customers are saying, well, sometimes my happy customers, I want to use the customer service product, but then they wanna buy something with the sales product. And so how can we, how can I as a, you know, somebody looking at uh technologies, make sure that the systems inter operate a lot more because I don't wanna have one system for this and then a whole other system for that. Um Just the way companies are doing business I think is really being disrupted. But even the way we're developing things at Microsoft, I can feel that myself, like who should I hire? How come this team which felt way over here before is now, you know, right here and how do we kind of move quickly and get everybody to coordinate better? You know, I think one of the things that I've been hearing just from talking with people in this room earlier today is that their companies are still not sure whether they wanna train the models from scratch, whether they wanna build or buy. What advice are you giving people these days? Yeah. So I run a product which is called Co pilot studio. So that lets people create their own co pilot. And what we're advising companies is maybe start with a tool that, um, is more of a fast service that you can get up and running really quickly and try with your customers. And then as you need to extend that you can go to building your own um, system. And one of the reasons I think we're advising companies to do that is, you know, it's hard to hire A I talent and a lot of times by the time you build something, the system has changed and then you're kind of reworking it. Um So I think that would probably be my advice is unless you have a really big A I team, I would probably start with some of the more pre built services and then customize that as you need because you'll get, you know, as we update things, you'll just get that for free, you know, that makes sense. Um Now we talked a lot also about company culture and branding and trust. Trust being a big thing, company culture being huge, especially if you're saying people talking about skilling, hiring upskilling. How are you advising people on how to brand and company culture intact while shifting to automation? Um Both from an employee perspective and a cons consumer experience. What um how are you building A I in a way that builds trust rather than breaks it down? I think that's, there's kind of a couple parts to that. I mean, first of all, security, I mean, someone here we were talking about security, um, security and a, I, I mean, there's a, there's so much happening, I think on the security front and so you definitely wanna make sure that your systems are secure. That's kind of the foundation of trust, I think. And then there's, you know, just, um, you know, there's kind of brand honesty, which might kind of be a kind of a different variation of trust, you know. So I think that um with security, you know, that's even with all our focus on A I that would probab, that would definitely be our company's number one focus. Um Just because we see so many bad actors using A I as well as good actors. So, um just a lot of focus there. And again, I think that's kind of why I would recommend people, you know, you don't want to build all of the systems yourself because there's just a lot of work that goes into um managing secure system. And then as far as, um you know, kind of maybe on the other side, one of the things that we see a lot of people doing is wanting to have an agent or an assistant um at your front door of your company. So, you know, in the old days you had your website, um or maybe the front door of your company, which, you know, I think people really think of that as anymore. So you might have, you know, an assistant that's out there um helping people ask questions or answer questions and just what is the persona of that? How does it build trust? How does it keep, you know, data secure and things like that? You know, how are those experiences, engaging and meaningful to people? So there's just a lot of different aspects from low level security to um privacy and responsible A I to just the way your systems interacting, you mentioned bad actors. And I wanna talk about that because I think that's a lot, there's a lot of scariness out there. What are some of the, the biggest examples of, of bad actors or any pitfalls you really see for relying heavily on A I? Mm I mean, there's probably a couple different levels. I mean, the probably the thing that you hear a lot about with um consumer systems is just, you know, what are you doing with my data? That's the thing that we hear the most from companies. Like, are you leaking my data out to a model? Um Are the things that my customers say? Does that go into making a model smarter? So like at Microsoft, one of the first things that we did is we took the open A I model and we definitely um made it so that companies could upload their own proprietary information and their customer data and keep that separate from the public models and that, that combination I think has been really great for people and really critical um for us, like, you know, we manage companies, email data and things like that. So obviously, if we were looking at your data or using it, like people wouldn't, wouldn't use our tools. So that's just been a really foundational piece of um of the work that we've been focusing on. So I love that you have guard rails around that. How do you advise companies that are trying to make these, these guardrails and these regulations and rules um within how they look at A I, how are you thinking about that at Microsoft? Well, I think what we would do is again, it kind of depends on what level of the service you do. So if you use the fast services, which is kind of what I was recommending earlier. A lot of those are kind of built into the system. So, you know what, we would allow an administrator maybe to control if you were using these services inside your company, like we're definitely um you know, letting admin control that. And then as you kind of go down the stack, you can customize that more yourself. And I know this dinner is about Ja I and we're going to talk about that, but what is, what is next, what's up and coming, what's on your radar? Um I think the thing that I would really love to see are more systems like the system that we built with github, which is, you know, where we're really seeing real productivity um and job satisfaction coming from the people that use the system. So we're in github, we're seeing that both from the programmers using the system as well as their managers. And like I said before, for like easy task that you would just want to do more repetitive tasks, you wanna do more easily and a more sophisticated task. Github is kind of unique because it has a lot of public data and people have kind of put a lot of their work history into a system already. So you're not starting fresh. Um We're starting to see that also with other verticals like customer support or um for a whole company, we're starting to see that where you have all your documents and your email, your meetings and things like that. How do you, how do you really or maybe let me take a step back. Like I think the advantage of gen A I is that it can deal with conversation data, a very unstructured information that you couldn't do before. So, you know, I worked on conversational ad when you kind of had to manually do a lot of this. And so it's kind of a miracle in a way that it exists and it's just been a crazy last two years. But I think a benefit for a company is taking all that information that was very unstructured, either to do a sales call or to just run a meeting in your company and be able to take to get more information out of that so people can be more productive. And so we've just been focused on that. It's, it's, it's, it's been great to see, but we're definitely just starting. Yeah, I know. And there's so much more to do. I feel like um and as I understand it's going so fast. So thank you so much Lily and we will leave time for questions for Lily once we said. So. Thank you. Thank you. Thank you.