In 2024, we organized the AI in Manufacturing Summit in Davos and Zurich. While I intended to summarize my learnings instantly, I am finally hitting the keyboard twelve months later… It seems human procrastination is a problem AI has yet to solve. Well, because a year is a decade in the AI space, it allows checking some of our predictions!
AI in Manufacturing Summit
For the Summit, we literally filled a bus with impressive individuals and drove back and forth between ETH Zurich and Davos during the World Economic Forum’s Annual Meeting 2024. A personal highlight of the Summit was a panel discussion on AI and the Future Factory in the AI House Davos.
The panel featured four luminaries in the area of AI in manufacturing: Eric Enselme from the World Economic Forum, Prof. Alexandra Brintrup from the University of Cambridge, Dr. Gunter Beitinger from Siemens, and Dr. Julian Senoner from EthonAI. Let’s revisit five of the questions discussed:
- Is this time different?
- Will AI enable the lights-out factories at scale this decade?
- Are manufacturers ready to benefit from AI?
- Are people on the front line of manufacturing better or worse off with AI?
- Is there a risk that manufacturers become overly dependent on AI and lose critical skills?
Keeping up with AI
Everyone on the panel and almost everyone in the audience agreed that this time is indeed different. The AI summer is hotter than ever. The speed of breakthroughs in computer science is challenging to keep up with, even for the people who work at the forefront of it. Among the Summit participants, two out of three companies thought they were lagging behind the average (even if they represented progressive firms). Good news for them; the others are struggling to keep up too!
The panel and audience had mixed opinions on whether AI will help realize fully automated lights-out factories within this decade. The industry experts doubted that self-aware, autonomous factories are around the corner, but they shared examples of lighthouse companies showing impressive progress. Perhaps they would see this question more positively today, given all the recent buzz about AI agents. Some say AI agents will soon close the automation gap currently covered by humans.
According to Amara’s law, we always overestimate what new technologies can do in the short term and underestimate what it does in the long run. However, a good question is whether a decade today is short or long run?
Gaps in the pack
Most manufacturers are still not ready to fully embrace AI. They lack the data and IT infrastructure to put AI to meaningful and broad use. This is especially true for small and medium-sized enterprises, which make up much more than 90% of companies worldwide. In a survey we conducted in the Swiss high-tech manufacturing sector in the spring of 2024, only three of more than 200 companies reported scaled AI implementation in their manufacturing operations! I expect this sobering low number to increase rapidly.
There are notable industry differences in AI implementation. Large companies in automotive, electronics, pharmaceuticals, petrochemicals, semiconductors, and fast-moving consumer goods often operate automated production lines, which makes them receptive to AI solutions. Gunter Beitinger from Siemens and Julian Senoner from EthonAI provided concrete examples of companies already benefiting massively from AI solutions, such as visual quality inspection, root cause analysis, and production scheduling and control. AI is also already common in forecasting and route optimization, according to Alexandra Brintrup. Eric Enselme mentioned that 80% of the companies awarded in the Global Lighthouse Network in 2023 boasted scaled AI solutions. By early 2025, that percentage has increased to 100%!
The real agents are human
On the question of whether AI will improve or destroy manufacturing jobs, the panel and audience were surprisingly positive. There seemed to be agreement that AI can help make better jobs—at least for the ones who will keep their jobs. The hope is that AI helps accelerate the automation of repetitive and tedious tasks and augments the workers in creative and relational tasks where human ingenuity and trust are needed.
Julian Senoner explained how EthonAI pioneered the Manufacturing Analytics System (MAS), a software platform that augments rather than replaces the human expert. The goal is not to implement AI technology per se but to improve the business. AI should assist people in solving the problems that matter to their business, also referred to as Augmented Intelligence.
The panel and participants also warned that manufacturers should not become overly reliant on AI. They prefer AI-in-the-loop or human-in-the-loop solutions to fully autonomous AI systems. This shifts the challenge from technology development to talent acquisition and upskilling. Workers who embrace AI tools but do not let them infringe on their domain expertise and human skillset will be in high demand. If anything, this issue is only amplified today, according to the Future of Jobs Report 2025. People matter more than AI!
What’s next?
What’s next for AI in manufacturing? The ground truth will be found on manufacturing shop floors and supply chains worldwide. Meanwhile, we continue the discussions in Davos during WEF’s Annual Meeting 2025. Visit the AI House for our new panel Augmented Intelligence: Taking Manufacturing to the Next Level, on January 22, 2025, or watch it on YouTube shortly after.
The AI House Davos panel debate from Davos 2024 is available here:
Leave a Reply