Artificial intelligence noun The theory and development of computer systems able to perform tasks normally requiring human intelligence

Can AI design or engineer cars better than humans? Jesse Crosse investigates

- Jesse Crosse

Like it or not, artificial intelligence (AI) has arrived and is very likely here to stay. It sneaked into the public eye almost unnoticed and, on the one hand, it appears as a harmless plaything that can give more informed answers to questions than a search engine can. At the other extreme, there's concern that AI poses a threat to humankind on many levels beyond the obvious prospect of robots laying waste to the planet. Somewhere in between, though, AI is already proving its value in engineering for making cars better and safer.

So if AI can be both a threat and an invaluable tool, what exactly is it? Richard Ahlfeld founded AI software development firm Monolith in 2016 as a spin-off from his PhD in aerospace engineering at Imperial College. "There are many different things that fall into the category of artificial intelligence, whether it's GPT models or autonomous systems or autonomous cars," he says. "I think what Monolith does is probably the most sensible form of AI. It's the ability to learn from data, allowing you to understand something so complex that normally you wouldn't be able to understand it." When trying to assimilate masses of information manually, the more data there is, the greater the task. With AI, the more data it's given, the better the results are. "That makes machine learning from data a fantastic shortcut through the jungle of complex things," he adds.

Ahlfeld became even more convinced of this in the early days of his company during a collaboration with McLaren to speed up development cycles using AI. By training AI on the physics influencing the car from data captured driving a car on track, AI was able to predict the results of testing. "That means test less. You can essentially condense six weeks of testing into two weeks," he says.