🎦 PEER: #AI ASSISTANT - video introduction A video featuring interviews and recorded footage from the #PEERproject kick-off meeting held at #VUB @vubrussel is available. more 👉 peer-ai.eu/ Subscribe to our #newsletter! youtube.com/watch?v=kaF97A_u…

Explainable AI 🔍 Explaining an AI decision is only the first step. Users also need to understand the reasoning behind recommendations and explore alternatives. PEER is developing a multimodal Explainable AI (XAI) approach that combines: ▪️ Natural-language interaction ▪️ Interactive visualisations ▪️ Adaptive explanations By making AI reasoning more accessible and interactive, the aim is to support users in understanding and evaluating AI-generated recommendations. Transparency becomes more meaningful when people can interact with it.

Sequential Decision-Making 🤖 Complex problems rarely have a single, definitive answer. They require a sequence of decisions, each influenced by new information, constraints and human preferences. Traditional AI systems may struggle to adapt to this evolving process. PEER explores a different approach: enabling users and AI to discuss alternatives, exchange feedback and refine decisions together. The goal is not just to generate an answer, but to support better-informed decisions throughout the process. AI should help us navigate decisions, not simply deliver conclusions.

September 23rd and 24th, the @CATIE_AQ team, represented by Joseph Gardette, visited CEA-List's offices in Paris-Saclay for the #ESAIM (4th European Symposium on Artificial Intelligence in Manufacturing) conference (lnkd.in/eQN4_m3E). Joseph presented the paper entitled: "Why and how to build trustworthy AI systems?". The paper presents the user-centred design methodology used in the #PEERproject to build trustworthy #AI assistants. PEER partners: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global

So what matters for the rest of 2026? If the first half taught us anything, it’s this: 👉 The AI gap is growing. Not between companies that use AI But between those who use it well, and those who don’t. Moving forward, the focus is clear: • Better use cases • Better product thinking • Better execution AI is no longer the advantage. How you use it is. #AI #PEERAI

Not all AI use cases are worth building || One of the biggest lessons from 2026: 👉 Just because you can build it, doesn’t mean you should The most effective AI applications we’ve seen: • Solve a clear, repetitive problem • Save time or reduce friction • Integrate seamlessly into existing workflows The worst ones? They look impressive. but change nothing. AI should create leverage. Not just attention. #AI #UseCases #Innovation

🔗 AI products don’t fail because of AI || They fail because of everything around it. What we’ve seen in 2026: → Great models, poor UX → Powerful tech, unclear use cases → Smart features, no real adoption Building AI today means: • Designing for humans first • Making complexity invisible • Focusing on outcomes, not capabilities Because users don’t care about your model. They care about what it does for them. #AIProduct #PEERAI

How can Explainable AI help us leave the car at home? 🤖 At the AISE 2026 workshop at IJCAI-ECAI 2026, Dr. Hendrik Baier gave a keynote exploring how Explainable AI (XAI) can contribute to more sustainable and accessible urban mobility. His talk, “Explainable AI for Sequential Decision Making, or How to Leave the Car at Home,” focused on a key challenge: AI systems are already highly capable of solving complex optimisation, planning and scheduling problems. The real challenge is often getting people to trust, understand and actually use the solutions they provide. In the context of urban mobility, Hendrik explored how more flexible and personalised XAI could help people better understand routing recommendations and make cities more accessible. One of the two examples presented during the keynote was the Amsterdam use case from the PEER project, which explores how AI can support people in navigating the city according to their individual needs and mobility requirements. The example highlights an important aspect of human-centred AI: a route is not necessarily “better” simply because it is the fastest or most efficient one. Personal preferences, accessibility, obstacles and individual circumstances can fundamentally change what makes a route useful. By making AI recommendations more understandable, explainable and open to interaction, XAI can help create a dialogue between people and AI, rather than simply providing an answer. The Amsterdam example was presented alongside a second use case addressing XAI for paratransit planning, where operators need to schedule and combine trip requests across a vehicle fleet. Together, the two examples demonstrate how explainability can support human-AI collaboration at very different scales. Ultimately, the question is not only whether AI can find the optimal solution. It is whether AI can explain that solution well enough for people to trust it, adapt it to their needs, and actually put it into practice. Let's have a look for read more 👉🏻👉🏻 sites.google.com/view/aise20…

A successful edition of the XAI Workshop at IJCAI-ECAI 2026 🤖 || This year’s XAI Workshop at IJCAI-ECAI 2026 was a particularly successful edition of a workshop that has been bringing together researchers working on Explainable AI for several years. The numbers speak for themselves: 38 papers were accepted this year, compared with 13 in 2025, and between 40 and 60 participants joined the workshop across the two days. The programme brought together a rich mix of research presentations, keynote talks, posters and open discussion. Two keynote speakers offered particularly thought-provoking perspectives on the future of Explainable AI: 🔹 Prof. Vera Schmitt, with “The Case for Actionable Explainable AI” 🔹 Prof. Francesca Toni, General Chair of IJCAI-ECAI 2026, with “Explainable AI for Contestability” The workshop also featured a poster session, creating an opportunity for participants to exchange ideas and discuss ongoing research in a more informal setting. The event concluded with an open discussion that quickly became lively and engaging, a great reflection of the interest and diversity of perspectives within the XAI community. Hendrik Baier also contributed to the programme with three talks presenting research conducted together with his students, sharing some of their recent work and opening it up for discussion with the community. Overall, the workshop was a great opportunity to connect with researchers, exchange ideas and discuss where Explainable AI is heading, from actionable explanations and contestability to new approaches for making AI systems more transparent, useful and trustworthy. A big thank you to all the speakers, authors, participants and organisers who contributed to making this year’s edition such a success. More information about the workshop and its programme: 👉🏻 sites.google.com/view/xai-wo…

🚧 What building AI in 2026 really looks like || Halfway through the year, one thing is clear: AI is everywhere. But real impact is still rare. In the past months, we’ve seen: → More tools than ever → More noise than value → More companies experimenting, fewer scaling The difference? 👉 AI that works is not about features. It’s about solving real problems. Over the next few weeks, we’ll share what we’ve learned so far building AI products in 2026. If you’re working with AI — this is for you. #AI #ArtificialIntelligence

🎯 Halfway through 2026. Here’s what we’re seeing. || AI is moving fast. Faster than most teams can actually keep up with. Over the past months, we’ve been building, testing, shipping and learning a lot about what actually works. Not the hype. Not the demos. The real stuff. So over the next few weeks, we’ll break down what we’ve learned so far in 2026: → what’s working → what’s not → and what actually matters if you’re building with AI No fluff. Just what we’re seeing on the ground. #AI #PeerAI

From 8 to 10 July, the @CATIE_AQ team, represented by Elise Durnerin, took part in the 13th EPIQUE 2026 Conference organised by Arpège, hosted by ENSIBS (Université Bretagne Sud) in Lorient (France). During the conference, the team presented the European PEER AI project via a poster entitled ‘Designing trustworthy and comprehensible AI assistants’. @CATIE_AQ won the audience award for the best poster. 🎉🎉 Congratulations to them and thank you to everyone who voted!

🌍 Explainable AI for Sequential Decision Making @ World XAI Conference 2026 || World Conference on eXplainable Artificial Intelligence We are pleased to share the successful outcome of the second edition of the "Explainable AI for Sequential Decision Making" special session, held during the World XAI Conference (XAI-26) in Fortaleza, Brazil. This initiative originated from one of the core research directions of the PEER Project, which explores sequential decision-making across a wide range of application domains. As AI systems increasingly support complex decision-making processes, ensuring that these systems are transparent, interpretable, and trustworthy is becoming essential for their real-world adoption. 📌 Session highlights: • Around 12 paper submissions were received. • Each submission was reviewed by approximately three Program Committee members. • Four high-quality papers were selected for presentation. • The session attracted greater interest and participation than its first edition at XAI-25, demonstrating the growing relevance of this research area. A sincere thank you to all the authors, reviewers, participants, and the session organizers: Ann Nowe, Mehrdad Asadi, Grzegorz J. Nalepa, and Hendrik Baier, whose commitment made this event a success. We look forward to building on this momentum and continuing to advance research on explainable, trustworthy AI for sequential decision-making.

How can AI help make cities more accessible for everyone? At the end of May, the @CATIE_AQ team conducted user evaluations in Amsterdam as part of the project's Inclusive City Navigation use case. Working closely with the City of Amsterdam and supported by local accessibility stakeholders, participants tested an AI-powered navigation assistant designed to help people with disabilities move more easily through urban environments. The solution proactively alerts users to potential obstacles and suggests alternative routes, leveraging city-wide data to provide a more accessible and user-friendly navigation experience. A special thanks to the City of Amsterdam team (Maike Ehlert, Eva Hageman, Vishruth Krishnan) for their invaluable support in recruiting participants and facilitating the evaluations. These activities are an important step towards creating smarter, more inclusive cities that work for everyone.

On 15,16,17 April, as part of the #PEERproject, the @CATIE_AQ team started the evaluation of the various MVP2 prototypes in SONAE’s offices. Over three days, the team conducted tests on the #prototype of an AI-driven personalized routing system plan with a panel of users to gather valuable insights and feedback. The @INESCTEC team was also on-site, offering technical support to ensure the smooth functioning of the simulation environment they had developed. A great collaborative effort to move the project forward! Frederico Borges Miguel Lunet Madelena Sottomayor Matilde Lopes 👉 Discover more 👉 peer-ai.eu/ And subscribe to our #newsletter!

On 20-21 April, the @CATIE_AQ team continued their work on the #PEERproject by advancing the evaluation of the MVP2 #prototypes. Najib Erbihi spent 2 days at Proditec’s office for the evaluation. Together with the team, they carried out a series of tests on the prototype of an #AI assistant designed to #support users in training an anomaly detection model for #coins. 👉 Discover more 👉 peer-ai.eu/ Thank you to all the partners that are making this possible: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global

🎉 On 7th May, as part of the #PEERproject, the @CATIE_AQ team (Elise Durnerin Charles Fage Aline Roc Joseph Gardette Enola Constanceau) finished the evaluation of the MVP2 prototypes. In collaboration with #Datacation (Floris Bleuland van Oordt, Joep Beurskens ), the team conducted a user test with logistic engineers on an AI assistant prototype used to improve the optimisation of warehouse operations. More info! 👉 peer-ai.eu/ Thank you to all the partners that are making this possible: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global

Another milestone for the #PEERproject 🤝 On May 6th and 7th, the RP2 Peer Review Meeting and the #PEER Consortium Meeting were held in #Brussels 🇪🇺, kindly hosted by VUB Vrije Universiteit Brussel. A huge thank you to #VUB for the hospitality. Over the two days, our partners presented the latest updates and participated in interactive discussions focused on key priorities, challenges, and next steps. It was an inspiring and productive gathering. We are proud of how far this project has come and excited about what's ahead. 💫 Stay tuned for all the results. More info! 👉 peer-ai.eu/ Thank you to all the partners that are making this possible: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global

POLICY BRIEF | Collaborating with AI users in early stages of AI development 👉🏼 Let's have a look: peer-ai.eu/outcomes/promotio…

🗣️ WEBINAR ALERT 📅 15 April | 11:00 CET How can we ensure that AI-driven public services remain transparent, accountable, and trustworthy? Join us for: 👉 “AI-Enabled Public Services: Building Resilience and Accountability” 👀 Discover insights from multiple research projects working on AI, governance, and public sector innovation. 🎙️ The webinar is jointly organised by the hashtag #HorizonEurope projects AI4REALNET Project, THEMIS 5.0, TANGO Project, HumAIne project, PEER AI, and AIXPERT. 👉 Register now and save your spot: lnkd.in/ewwquttP #TrustworthyAI #PublicSectorInnovation #ResponsibleAI Thank you to all the partners that are making this possible: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global

🗣️ WEBINAR ALERT 📅 15 April | 11:00 CET How can we ensure that AI-driven public services remain transparent, accountable, and trustworthy? Join us for: 👉 “AI-Enabled Public Services: Building Resilience and Accountability” 👀 Discover insights from multiple research projects working on AI, governance, and public sector innovation. 🎙️ The webinar is jointly organised by the hashtag #HorizonEurope projects AI4REALNET Project, THEMIS 5.0, TANGO Project, HumAIne project, PEER AI, and AIXPERT. 👉 Register now and save your spot: lnkd.in/ewwquttP #TrustworthyAI #PublicSectorInnovation #ResponsibleAI Thank you to all the partners that are making this possible: @VUBrussel @Fujitsu_DE @CATIE_AQ @INESCTEC @Eurecat_news @TUeindhoven @UniKarlova @JagiellonskiUni @SONAE @proditec @AmsterdamNL @Fujitsu_Global