Video transcript: Envision on-demand panel session - How human ingenuity and generative AI technologies can combine to deliver business value

Playback of this video is not currently available



Darshan Chandarana: Hello everyone, and thanks for joining us today. My name is Darshan Chandarana, and I'm the Head of Emerging Technology for PwC. I'm joined by three panelists today and we'll be talking about all things AI. I'm joined by Wendy Castairs who is focusing on strategic partner planning technologies around the AI space there. Joined by Julia How, who's focused on the workforce and the workforce experience, And joined by Leigh Bates, who looks after responsible AI in the FS sector. Wendy, there's a lot going on at the moment in the AI space. It's unprecedented in lots of ways. But you've made a big announcement. Microsoft A, launching something on 1 November. I'm not going to steal your thunder. What is it?

Wendy Carstairs: Microsoft 365 co Pilot is coming to market. It's a general availability on 1 November to be available to anybody to increase their productivity in the workspace. So frankly, Arjan, this is the biggest change to workforce software that I've seen since the inception of word wander back in 1983. It is going to change the way anybody that uses a computer to do their work gets the work done. So it's so exciting. So we're drawing on the power of the large language models and open AI. So that capability in chat GBT in an enterprise way that can allow us to interact with our information and our data. A jobs done. My favorite example at the moment, because I've had a bit of an access to it, is Teams meetings. I'm on a Teams meeting at home. Doorbell, goes, courier is there with a parcel. And I get back to the meeting and I'm pretty sure I've missed something significant, but I'm not sure what. In private, I use copilot to create me a summary of what's happened in the meeting so far. Who said what? What documents were referred to? Timestamps and anything that might have happened within seconds. I can quickly look at what's going on in the meeting so far and then catch up. And in 5 minutes before the meeting, I can ask co pilot to give me a summary of the whole meeting. Kind of reflect, have a think of what do I need to actually follow up on. Maybe there's some questions I need to ask right now because I haven't absorbed those the first time around. And then at the end of the meeting, everybody gets the summary. If you want one action points, who said that? They're going to do what? And then we've got that as a follow up in the meeting chat for me that's game changing on its own. But then of course doesn't stop there, does it? All the office tools are going to be enabled to co pilot Excel. I'm a huge user of Excel, but not a very good one. I've been using it into the star and I know how to get the basics done, but I always find myself wanting to just get a bit more information, some insights out of all these lists that I create. I found that copilot can just really help me analyze it, interact with the data, and I can get some questions answered. It teaches me as well because it shows me how to go about those, all those functions and things that I never had a clue how to use before. It can show me how to do those in seconds. I can get answers about the data. And then who isn't a user of Powerpoint nowadays? Powerpoint within seconds, co pilot can create for me a summary of all of my data and all of my big documents I asked to summarize up. It provided it with a lovely bulleted Powerpoint three slides, ready for the exact meeting. I and I can pull it together with a lovely design just really fast using my data and my information that's within our enterprise knowledge base. I think you'll find there'll be a lot of takers up of the Powerpoint side of things on our side of the fence.

Darshan: Anyway, you've mentioned the office side of things and you've just touched on open AI as well. Give us a little bit of news about that. What's going on with that partnership and where are you taking that as well?

Wendy: Well, of course, since the announcement of open AI services to be enabled and available in Azure for anybody to create their own integrations into your business tools and business software. The demand has been unprecedented. So to be able to access the Chat GPT, the GPT 3.5 and now GPT four models within Azure in a way that is enterprise class, that is using your data in a completely protected and private way, as well as within a responsible framework. The demand has been unprecedented. It's been the fastest growing service in Azure since Azure was first built. So we've been building our capacity extremely quickly. I've never seen it built so fast. And now anybody can access these models to create your own tools and your own software. And people are moving from proof of concept into production. Right now, we've already got over 4,000 customers who are running production services using open AI, large language models in an enterprise class and safe way.

Darshan: That's fundamentally going to change the way that we interact with our tooling and the way that we interact with the office suite. Julia, what does that mean from a workforce standpoint? How is that going to affect the way that we interact on a day to day basis with ourselves and with our colleagues?

Julia Howes: Yeah, I would say all of the organizations that we're looking at and working with at the moment on AI, there's I guess, a common thread between those organizations that are really leading. And that common thread is really focusing on unleashing the energy of their employees around this. And that's a challenge because AI and the generative AI has got a number of unique aspects to normal change in an organization for employees, It's very personal. They take it home with them every day at these concept. That's not something that's unique to the workplace anymore for a lot of people. That can create mistrust and fear for other employees. That can create unbridled enthusiasm which can be unchecked. Also, there's a bit of a curve of disillusionment and disappointment. There's so much anticipation around it, but if it's not properly used, it can lack the initial impetus. There's a lot for organizations to unpack. We've been talking to them about this principle of 80, 20. So if they spend 20% of their time on the actual algorithms and data, they should be spending 80% of their time on the business and in people transformation around it. And that's where we see leading organizations really focused on the employee areas. There's, I think, a baseline of actually creating that trust and giving them a voice in the AI and the use cases and the agenda. There's a lot around reskilling, but also giving the individuals personalized learning approaches and giving them the time to upskill properly. Then there's a large piece around the actual business itself. I think what's really game changing about AI is it will totally change how organizations need to interact. So there's a big piece around culture and ways of working, how teams work together around these things. That if that's not unlocked, then I think the potential of AI is hard to imagine. And then finally, there's a lot around responsible AI and the ethics of that. And at a personal level, that's very important for employees to understand and feel comfortable in the organization's approach to that. I think where we've been working, we're seeing some really exciting opportunities around this is actually using the Microsoft Suite, so using AI to help with AI adoption. So we've been working with Microsoft Fever in particular to help generate very personalized change journeys rather than one approach fits all. It's really understanding both the role impacts of AI, but also the individual mindsets of the person and how they're going to come to the AI journey. To then create the right content at the right time in the flow of work, but also to monitor and adapt to how employees are working through their agenda. Then it can continue to push. If there's any dips in that change journey, we're seeing a very exciting advent in this complex, but I think, you know, the technology is there to help us with it. I want to take out one point of that and actually pass that onto the responsible AI piece. Actually, responsible AI is something that we're quite serious about here at PWC and with the government's recent announcement as well about the AI summit, the first ever I think.

Darshan: Yeah, what's our position on that? Where do you see that going?

Leigh Bates: Yeah, it's a great question. And look, I think you know what an exciting time we're in right now. We've heard about the pace of technology change in this space is just fantastic to see. But with generative AI, obviously using large language models, there come some risks. Because it's not that easy to explain exactly what a language model is producing. It can hallucinate. We've heard this term quite a lot. And therefore, AI, safety and responsible AI and ethics has risen right up the agenda for many conversations we have with boards and government, et cetera. And hence, the UK government really wants to take a leading position in this space, right, with the AI summit focusing on safety at PwC. Like I think you mentioned it, right? It's not new for us. We've been involved in responsible AI for many, many years. What's changing now is really thinking about the holistic framework to ensure it's sustainable. And you've got the right controls and guardrails in place across the broader state of AI, not just generative AI. So mathematical models as well. What I would say is that there may be a bit of a myth out there that responsible AI, ethical AI is putting the brakes on innovation. It's stopping innovation. That's definitely not the case, right? It's not there just to identify when a model is hallucinating or when you've got model drift and stopping that particular process. It's much more of a holistic framework. Think about all the things you need to have in place around the control environment, the governance environment, identifying areas of security risks in the cyberspace, privacy, IP protection. All these things that you need to think about as you think about your AI journey and really adopting a responsible AI approach. Right at the center of your AI strategy. At PwC, we have this saying, Darshan, of being human Led but tech powered, there's nothing more relevant in this space than that's saying really. Because what we're really seeing is that you need to make sure that you've got the right individuals, the right humans in the loop, when you've got tools like we've been talking about today, to ensure that you've got the right guardrails and you're using it in the appropriate way. But ultimately this all comes back down to value, right? How are you going to drive value from the investments in AI? Maybe I'll throw it back to you, Darshan, to talk a little bit about, you know, what, some of the things we're seeing across our clients and how we're helping them select the right use cases.

Darshan: And I think that is the big question in the market at the moment. There's so many tools and technologies out there. The pace of changes, as everyone has mentioned, is just something I've never seen in the industry before. I think what we'll probably see in the next few months is a sort of a coalition of different types of models coming in. So one model won't answer everything. That's one thing that we're starting to see. So you might need a multi model approach. The other piece that we'll see is there'll be things that will be rolled out from Microsoft for example. Which will just be in the general population within the entire organization. And people get used to it, people get used to working with it and comfortable with it. And then that will change the way that the workforce actually interacts with AI. More ideas will start coming out and so on. Coming back to your point about sort of where do we see the use cases. I think in a regulated industry, there's going to be a lot of work we'll have to do with both the risk teams, government agencies, the FCA, and others to make sure that we're progressing in the right sort of way. But also picking the right model for the right use case. It's not a one and done approach. We'll see evolution over time. Look, I think that's where we're going. It's been a great discussion and thank you all for your time here today. All I can say is over the next six months we shall see even more change. I'm still reminded that chat GPT only came out in November last year, so it hasn't even been a year yet. Who knows what the next year will bring. But we'll be there with our clients to advise and help. Thank you.

Follow us