Feature

How much artificial intelligence is integrated in machines nowadays?

Since ChatGPT, artificial intelligence has been unmissable from the media. But what does it actually look like in the industry? To what extent is it already being applied there to make smarter machines? Here, Dr Johannes Kühn, Head of Control Engineering, Motion Control and Robotics at Lenze, explains where we are today.

29/05/2024 17:06:07

Font size:

"In production itself, in the heart of the machines, there is no artificial intelligence integrated yet," Kühn immediately tempers expectations. "That doesn't surprise me so much yet. Reliability and continuity are the
keywords there, there is no room for experimentation because in themselves, there is not that much real intelligence in the methods we apply in industry today.

“Machine learning, which gets the most application
in industry, is one of the weaker methods to apply and has been around since the 1970s. It boils down to pure statistics: classifying data and making decisions based on historical data.

“The reason it is coming to the surface more now is that there is enough computing power to make those predictions, but you can train a model so well that if a completely
new situation then arises, it won't have an appropriate response in-house. A machine builder cannot afford that. That's also, for example, where the self-driving car encounters its limits."

Read the full article in DPA's June 2024 issue

Written By Sophia Bell

Sophia Bell is the Group Editor of DPA, Connectivity, and PBSI. She joined the team in 2020 as Assistant Editor after achieving a First-Class Honours BA in English Literature and Film Studies from the University of East Anglia. In her current role, she leads the editorial strategy, driving digital growth and ensuring each title continues to deliver trusted, relevant insight for engineering and manufacturing professionals.