The emergence of generative AI is revolutionizing the business landscape, offering unprecedented opportunities for innovation, efficiency, and growth. As more companies adopt this transformative technology, it is crucial to recognize both its immense potential and the new challenges it presents.
At the 2023 G1 Global Conference, Microsoft Japan President Miki Tsukasa joined panelists Stephen Barnham, Sinead Kaiya, and Yoshinami Takahashi to discuss how companies are taking advantage of the opportunities brought on by generative AI.
How Generative AI is Improving Driver Safety
Sinead Kaiya: On the automotive side, a lot of the investment is going into what we call ADAS or Advanced Driver Assistance Systems.
As you can imagine, this has been years of using machine learning to train your car, to be able to sense the environment that it’s in, predict what might happen next, and then plan a course of action.
This is extremely complicated and has been developing over many years, but one of the challenges or bottlenecks of development is trying to expose the models to enough real-life scenarios, especially edge cases, to improve safety.
To give you a concrete example, one of the things that these companies can do is pull the real camera data from actual cars that are on the road. So show me all the images of all the cars that were on the road yesterday. Use AI to search for what we would call near-miss accidents. Then they can take the imagery of these near-miss accidents and use generative AI to create new simulations. So for example, a car nearly hitting a bicycle. It can show [what happens if] the bicycle going faster or slower. It can show me the scene in the snow, or the rain, or at night. You get the picture, right?
These are scenarios that we would never be able to recreate in real life that engineers can now use to train their models to make the cars safer. And what’s fascinating, and maybe what business leaders across all industries might want to think of, is the car now becomes a product that gets smarter over time, and it gets safer over time.
Your car used to be a product that you bought, and slowly it would degrade and wear down and it would get less safe over time. But now, by pushing these updates to the car regularly, the product increases in value to the consumer over time. That is a business case that nobody can argue with, but it’s also one that is, of course, incredibly good for society in terms of reducing traffic fatalities.
Adopting Generative AI in Human-Centered Industries
Stephen Barnham: When I started in the technology industry, I started in defense. And in those days technology flowed from the top down. It started with governments who had big mainframes. They were the only organizations that could afford them.
Then the minicomputer came along and corporates adopted the technology. And then IBM came along with the desktop running Microsoft software. Then came the democratizing effect of consumer software. So it was very much top-down.
Now, largely driven by smartphones, the abundance of data, and the internet, it works the other way. My 88-year-old father has arthritis. He can’t [physically] operate a smartphone, but he does it entirely by voice. And he uses chatGPT by voice to write poems for his granddaughter, who’s just graduating. In this way, consumers are testing out technology before corporations and governments can. That’s because corporations have cyber and regulatory concerns, and governments have all kinds of bureaucracy. So society has reversed because of this massive uptick in in computing.
At Dai-ichi, we’ve been thinking about that. And rather than everything being the job of some top-down chief digital officer like me, we’re trying to democratize digital and democratize AI using tools such as Sandbox that we have created in Microsoft Azure.
We’ve pushed this out to digital communities. We have over 50,000 staff members in Japan, and within these sandboxes, they can experiment. They can just go crazy. It’s completely safe. They can import their data, and they can test out things like analysis of customer feedback or prepare smart systems to support the sales conversation.
We are also experimenting with chatbots with avatars and clearly, they’ve come a long way. They are very human-like. However, [in the life insurance industry] I can’t see a way in the near future that a human would be taken fully out of that loop. And really, our customer service agents are such fantastic professionals. They do a very difficult job every day. I’ve been a technologist for close to four decades now, and I deeply believe in the power of technology. However, I think insurance is a hybrid business.
Using AI to Run Your Business More Efficiently
Yoshinami Takahashi: First of all, let me start from the positive. So I think within Fujitsu there’s a lot of proposition that we provide to the customers. We’ve been an AI company for over 30 years. So we have wide learning. We have a deep learning. So we’ve been pretty much acquainted with AI.
I think within generative AI there could be a big band, especially when we talk about [moving from] search by product to search by necessity. I think this is one of the biggest door openers—consumers not searching for the product, but generative AI defining the hidden needs of the customer.
So I think the world is going to change a lot. Especially in Fujitsu, we are providing demand prediction, especially around the retail sector. And we think providing demand prediction is going to improve the supply chain as a whole. So making sure that the retailers are not oversupplying. This is going to be very important to minimize food loss.
Another point of how we can better improve within Fujitsu is data integration. We have over 50 fragmented systems, which have been aggregated into one big data lake. So now within Fujitsu, we are using a dashboard to make our decision more effectively. By doing so, we minimize the time spent on nemawashi.
So again, through the usage of generative AI, I think we can bring this a step forward, especially in the area of internal documentation. I think we’re spending way too much [time on] documentation. By using generative AI, I think we can minimize the document creation.