Industrial AI projects are challenging.
Unlike AI projects in the virtual world where lots of data is typically available from e-commerce, social media etc. data in industrial is highly individual, often noisy, and also often not available in a usable form.
Then, while in the virtual world a 70-80% accuracy would be highly business relevant, in industrial 90+ % is often needed. Also, safety and security issues are highly relevant in industrial processing environments.
In this talk and interview we address these questions and why both sides, AI and domain experts are needed to run a successful project.