Speaker Notes Peter Seeberg to EFFRA EU Commission for workshop on AI in manufacturing in Brussels
1. “AI Potential and Barriers in Mechanical Engineering and Manufacturing SMEs”
My pledge is going to be towards Embedding ML for Continued Global Product Leadership
And we may only be able strengthen our number One global position through Education
Peter Seeberg, AI Consultant for asimovero.ai - Member of Machine Learning Expert Group of VDMA
LSTM - Sepp Hochreiter, Jürgen Schmidhuber 1997 - Article Hannover Messe Industrial Pioneers
- „I'm worried, dear mechanical engineers!“ „ Do not mess up the lead in plant engineering.“ „We should not follow
Google or Baidu.“
- „But focus on AI in mechanical and plant engineering.“ „The talking machine must be our goal, not a new
smartphone.“
- The user speaks with the machine: "Watch out for speed.“ „No problem, "replies the machine," I got new special
oil, so I can drive faster.“
Mechanical Engineering AI Use Cases
(BTW: We talk AI but really mean ML!) (Used all around us!) (ATM, social media, …, …) (Best example: radiology)
1) Improving Overall Equipment Efficiency – Quality- Throughput – Standstill (Today 75% Predictive Maintenance)
2) New Business Models: PAY PER USE (Instead of selling machines, sell the good to be delivered, e.g. air, and
measure delivery through data.
Results from a VDMA Survey from early this year show us the following main areas: (ML implemented / to be
implemented (Customer Support 10% / 50% - Development 10% / 30% - Production 10% / 30%)
So far… POTENTIAL… Moving to BARRIERS
After you have decided you want to put embedded ML into your machines… How to move forward? Build or Buy? In
many cases you have DATA (Which so far you have not used for data science but for tracing product charges) In most
cases you do not have Data Scientist
Results from a VDMA Survey from early this year tell us the following reasons for NOT having had introduced ML
based solutions: Lack of personnel (60%), Lack of qualified data (60%), Lack of qualified personnel (50% ), Unclear
Use and ROI (40%), Legal questions (30%)
Combined I would say Lack of Qualification. That’s why last week here in Brussels at the European AI Alliance where
the POLICY AND INVESTMENT RECOMMENDATIONS FOR TRUSTWORTHY AI was handed over to Commissioner Gabriel with:
1.1. Encourage Member States to increase digital literacy through courses (e.g. MOOCs) across Europe providing
elementary AI training.
VDMA has been doing a lot; actually the ML Expert Group is putting together an overview of MOOC’s for self-
learning and organizing face to face introductions
I personally believe that every European should get the right to receive a 60 minute introduction into AI (for 1.
Understanding it is not an “I. Robot” dystopia 2. Understanding it will change the face of the earth 3. Deciding what
that means for one self.) Also, a partner and I have a weekly podcast “KI in der Industrie” targeted at decision
makers in industry planning to introduce ML into their machines and factories.
Conclusion:
1) „We should not follow Google or Baidu. But focus on AI in mechanical and plant engineering.
2) We need to educate Europe! Private persons as well as companies, in our case those of mechanical and plant
engineering.
Thank you!