Note
AI at the Danube conference focused on industrial AI, bringing decision-makers and contributors together to discuss topics like explainable AI, user experience, and automation. The event highlighted the importance of finding the right tools for the right problems, avoiding complex solutions for simple issues, and sharing lessons from failed projects to learn and improve. Future events are planned in the Alps and possibly Eindhoven, with a focus on small group discussions without PowerPoint, fostering collaboration and knowledge sharing in the industrial AI community.
Highlights
-
2024-10-24 15:38 I believe I said Linz. And I must say. And of course we’ll talk about the details of our conference as well. Linz, a beautiful city in Austria. AI powerhouse. AI powerhouse, right? We learned a lot compared maybe to what Eindhoven is for the Netherlands is maybe Linz for Austria as far as AI is concerned. Yeah, I think so. I was surprised that, I think 5000 jobs in software development and AI in Linz, that’s huge, right, for a city like Linz.
-
2024-10-24 15:40 Another topic was edge device and the associated topic of inference without Nvidia. That was written I think three or five times. And I read today Nvidia is now also building its own LLM. So yeah, I saw that. Yeah. This worries the others. They’re using the meta, some kind of lama, right? Yeah, but industry is looking for other solutions, as you mentioned, XLSDM could be there, but you have more than that. And that’s where Jan Kutnik came into play, combining Genai. He’s not a real big fan of Genai and reinforcement learning
-
2024-10-24 15:38 And she had this famous quote which I forgot, but it’s the same idea, let’s see that the robots do the dirty, the dirty difficult work for us, so we can do the wonderful work which has to do with creativity in playing football. And that’s why the quote was there. What did it say? You know, make sure that the robot’s automation does the work for us. Yeah, that’s the idea, yeah. Do the hard work, the dirty stuff. Yeah, sure. And the other thing that was under his name and the blue collar is, was in between it. No genii, no channel.
-
2024-10-24 15:39 . Yeah, because he said there are a lot more use cases on the shop floor. Please do not concentrate only on gen AI use cases. We can solve a lot more problems. And I have a quote from his session. Many companies find it difficult to distinguish between reinforcement learning approaches and optimization problems. They try to solve optimization problems with reinforcement learning that turn a simple problem into a big problem, often because there are tools for complex problems and that’s a problem for the companies. And said we need to find the right tool for the right problem, and then we can find a very good solution.