Everyone’s talking about AI – but who’s explaining it? 👀 That’s where Explainable AI, or #XAI, comes in. In our latest newsletter, we “explain” explainable #AI – showing how it turns the black box into a transparent, trustworthy partner across the entire lifecycle, from data prep to deployment. Learn how XAI builds trust, boosts safety, and helps teams make AI decisions for the right reasons. Also in this issue: • Breaking news on the world's first Industrial AI Cloud • More on the enterprise-operational AI Gap • Updates on the European Commission’s dual AI strategy Enjoy the read!
✨✨✨ What XAI really means is not machines explaining themselves — but humans finally understanding what they’ve created. ✨✨✨
Excellent article — very well written and focused on a highly relevant topic. I first became familiar with the Explainable AI (XAI) philosophy during my thesis project, completed a few months ago, and I truly consider it a fundamental and indispensable aspect in the development of high-risk AI models. An operator who is subject to the decisions of an AI model (just think of modern Decision Support Systems across almost every industry) must be able to understand why a given decision has been made. Moreover, thanks to XAI, it is possible to address and mitigate learning biases, which remain a major issue in today’s AI landscape. In conclusion, XAI represents a rapidly growing market that will gain increasing relevance across a wide range of sectors. However, I believe that the current techniques and tools available are still in their early stages and need further improvement to enhance both their practical applicability and their ease of implementation.
Carolina V.
2d
Well needed introspection on emerging enterprise AI systems with operational AI model and impressive developments Siemens always got us used to, well done ! And "why understanding how AI makes decisions is just as important as the decisions themselves" ! 👍🏻👍🏻🙂
I believe deep learning derives its value from recognising patterns that lead to certain outputs, and these are not patterns that a normal human can recognise. If you have a huge number of dimensions as input, it’s more often that not, not going to be possible to explain outcome probability to a human moderator
Explainability is what truly bridges innovation and trust. Without understanding the "why", even the most accurate AI remains a black box.
Hi sir I am interested
👍 On our R&D tracks on visual explinable AI, or XAI. https://www.linkedin.com/posts/cyrus-f-nourani-b7389419_a-generative-ai-computing-logic-request-activity-7093239984823922688-7T4C?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAPeKocBgu_Po3EIogrmi2KSmEfs1rFNJiU
Hallo Siemens, Die Entwicklungsabteilung von TIA Portal sollte diese Veröffentlichung lesen. Ebenso sollte sie vom Hauptverantwortlichen für die Entwicklung von TIA Portal gelesen werden. Siemens Forever! https://www.linkedin.com/posts/pablo-baena-vela-a83967332_tia-portal-544784-activity-7392980283651739648-UC-e?utm_source=share&utm_medium=member_desktop&rcm=ACoAAFPKp0MBzFx_wjJq-XtN3Y9QWZD-N57s9O8
Responsible AI Governance | Value-based Engineering, IEEE CertifAIEd, ISO 42001 | Engineer, Auditor
1dGreat article, I wonder if there is still a gap to close? If the „attention“ is off that explains wrong classification - so seeing „good attention“ in a model could certainly helps to build confidence in its performance. But when the model attention is bad, do we understand what’s the root cause? It’s a bit like asking a friend „Why are you not answering?“ and they answer „I don’t pay attention.“ - this answer would certainly trigger a follow up as to „why are you not paying attention.“