Artificial Intelligence (AI) is being talked about more and more by businesses, governments, and the public as the technology that’s going to revolutionise everything we do. We already see AI and machine learning used in several areas like within smart assistants, driverless cars, and various apps.

Despite the growing usage of AI in various sectors, integration of the technology into business processes is still very much in its early stages and we are a long way off living in an entirely AI-powered world.

“We’re in a pre-cursor mode, if you want,” said Charbel Aoun, AI city managing director EMEA at Nvidia, speaking at Unbound London. “It’s an early adopter situation because it’s new and it’s challenging because people have to understand what you’re doing to adopt. So, you have to go through a journey of proof of concept.”

A big hurdle that needs to be overcome before we reach the stage of ‘AI everywhere’ is the amount of unstructured data that exists behind AI, according to Fergus Weldon, director of data science at Trainline.

“A lot of the chat in the news is around new advances in academic research, algorithms, and methods but when you try to apply that in a business sense it’s a bit more Wild West. You don’t have highly structured data like audio or images. And when you try to mix different datasets together, problems actually become quite hard to solve,” said Weldon at the same event.

“In a commercial setting, it can be hard to scale AI, purely because datasets are constantly changing and, in a lot of the cases, you’re not using your own data, you’re using data from another service that you’re not responsible for.”

At the same time, with the rise of technology, the vast amounts of data are already being used to help drive certain platforms.

One such platform that has made use of all of its data sources is Google-owned navigation app Waze, which utilises AI technology in the backend of its app to aid make sure the frontend provides the necessary driving routes and its advertising business.

“Any company with a decent data source are doing certain things. There’s the pattern-recognition stuff – which we do a lot of and is ‘we think next quarter we’re going to have this amount of users, how are our systems going to be able to cope?’ and you can make predictions for that,” said Finlay Clark, UK country manager at Waze, at the event.

“We have an advertising business with thousands of advertisers and you can begin to predict which ones are likely to churn based on their behaviours… We can look at which groups of people are most likely to respond well to push notifications… All that stuff is machine learning 101.

“I think where it’s slightly interesting is where, for us, we have this huge traffic data source and we’re beginning to think about how we put new features in using machine learning. A good example would be motorcycle mode.”

Looking ahead, the three panellists predicted that we would see AI and machine learning help transportation to become more demand-driven, the focus of AI to fall within virtual reality and augmented reality on the back of use of deep learning, and the introduction of services such as kerbside delivery in the knowledge that people will be in a certain place at a certain time.