Industrial communication is a supreme discipline, and many of you will remember the standards discussions, the fieldbus wars, and the TSN issues. Industrial communication seemed to be solved. And now we have AI agents and many of you are wondering how to deal with them. Are they a new layer?
MCP and industrial communication
The AI world has been talking about MCP or A2A for weeks – the new "standards" when it comes to agents. But how do we actually introduce them to the industrial shop floor? Peter Seeberg from the Industrial AI podcast spoke with a legend in the field of industrial communications: Colin Masson of ARC Advisory Group. Our listening tip of the week.
Large language models tend to focus heavily on the first token in the sequence – creating a so-called attention sink. Many papers have studied this phenomenon in detail and proposed various ways to either exploit or mitigate it. Why do LLMs learn such patterns and how are they used? This week's paper tip comes from the island.
You can't do without arXiv. The global scientific community relies on it. We often link to interesting AI papers, and they are all on arXiv. But what is it? Wired has a very nice feature on arXiv and the people behind it. It is a long story, but well worth it for the deep insights and great photos.
The University of Amsterdam is one of Europe's AI beacons. Now they are offering a new Masterclass for AI Foundation Models. The masterclass will be held by Professor Cees Snoek, one of the leading researchers in the field of vision and AI. The masterclass is also aimed at employees in companies – it is not a purely academic event.
