Our PoC experiments show that this approach leads to logical alignment between answers and explanations, while improving their overall quality.
The core idea is that answers and explanations only extract information from the output of a reasoning process, which does not need to be human-readable. To improve faithfulness, explanations do not depend on answers, and vice versa.
I'm happy to speculate that our general technique for grounding explanations in LLM reasoning, presented at last week's XAI 2025 conference, could pave the way for finally cracking natural language explanations.
https://arxiv.org/abs/2503.11248
CRISIS IN MACHINE LEARNING - Semantics to the Rescue
Frank van Harmelen starts his keynote at ISWS 2025 with this headline from "The AI Times"
So, what is this crisis about? There are the following (still) unsoloved problems in AI research:
- Learning from small data
- Explainable AI
- Updating
- Learning by explaining
Portoroz coastline by night. Tomorrow, ESWC 2025 will start with 2 days of workshops & tutorials, before the main conference with the "official" opening on June 3rd. Looking forward to a great conference! :)
https://2025.eswc-conferences.org/about/
#eswc2025 #semanticweb #semweb #knowledgegraphs #reliableAI #explainableAI #llms #slovenia @fiz_karlsruhe @fizise @tabea @enorouzi @sashabruns
Can you spare 10–15 min for an #AI research project? My student is studying how explanations impact trust in AI decisions. No AI experience needed!
English: https://survey.trx.li/index.php/193548?lang=en
Italiano: https://survey.trx.li/index.php/193548?lang=it
RTs greatly appreciated! #ExplainableAI #Research #Survey
Applications are now open for the 2025 International Semantic Web Research Summer School - #ISWS2025
in Bertinoro, Italy, from June 8-14, 2025
Topic: Knowledge Graphs for Reliable AI
Application Deadline: March 25, 2025
Webpage: https://2025.semanticwebschool.org/
Great keynote speakers: Frank van Harmelen (VU), Natasha Noy (Google), Enrico Motta (KMI)
#semanticweb #knowledgegraphs #AI #generativeAI #responsibleAI #explainableAI #reliableAI @albertmeronyo @AxelPolleres @lysander07
Exciting news! I’ve co-authored two accepted contributions to #CHI2025!
XAIxArts Manifesto: Explainable AI for the Arts (alt.chi)
Sketching in HCI: Scribble, Sketch, Elaborate, Teach (course)
Befuddled by all the recent #DeepSeek hullabaloo? Here's a brief Q&A that cuts through the fog.
Q: Did #DeepSeek just up-end everything we know about #AImodels and #LLMs?
A: Nope. It just demonstrates one of several new approaches to model training and logic chaining, but still uses the same basic building blocks.
Q: Does this mean DeepSeek can think?
A: Nope. Still not #Skynet. Logic chains are just one of several techniques an instruction-oriented AI system can use to try to stay on track and focus on a coherent goal.
Q: Is logic chaining #ExplainableAI?
A: Nope. Even the "thinking" output of DeepSeek is a linguistic approximation of the pattern-seeking behavior of most LLMs.
Q: Why is everyone in an uproar about DeepSeek?
A: Because most people think ChatGPT defines what AI is, what it can do, and what its limitations are.
Q: Why are the people panicking about DeepSeek talking about AI hegemony and geopolitics?
A: Because they're more concerned with investment returns or charging for expensive GPUs and SaaS services than they are in scientific advancements or improving individual productivity with new technology.
Excited to attend my first PhD defence at TU Delft: Daniël Vos will be defending his thesis "Decision Tree Learning: Algorithms for Robust Prediction and Policy Optimization", containing work he did under supervision of Prof. Dr. Ir. R.L. Lagendijk and Dr. Ir. Sicco Verwer.
Share with us your research on #generativeAI and #KnowledgeGraphs!
Deadline approaching:
Workshop on Generative AI and Knowledge Graphs (GenAIK 2025), co-located with COLING 2025.
Submission deadline: 5 November 2024
Website: https://genetasefa.github.io/GenAIK2025/
@GenAIK @GenAsefa @fizise @fiz_karlsruhe @albertmeronyo #coling2025 #genaik2025 #AI #SemanticWeb #llms #explainableAI #robustAI #robustLLMs
Just received news that today, Semantic Web Company (SWC) and Ontotext have joined forces and merged into Graphwise! Definitely the right step to bring platform, tools, methodologies, and expertise all together.
Graphwise website: https://graphwise.ai/
The **BEWARE2024 Workshop** is happening at the **Associazione Italiana per l'Intelligenza Artificiale Conference** on **November 26th, 2024, in Bolzano**!
Programme of **accepted papers** for this year: https://sites.google.com/view/beware2024/accepted-papers
#BEWARE2024 #AIxIA2024 #AI #FairnessInAI #EthicalAI #TransparencyInAI #ExplainableAI #AIResearch #BiasInAI #AccountabilityInAI
This Friday at 12.15 CEST I'll be hosting a talk by computer scientist Kary Främling, in which he will present his work on Explainable AI techniques for producing explanations useful to a variety of stakeholders, rather than only to AI experts.
The talk is hybrid, so even if you are not currently enjoying the very colourful (and very wet) autumn in Umeå, you can join nonetheless!
More information here: https://www.umu.se/en/events/fraiday-social-explainable-ai-how-can-we-make-it-happen-_11985341/
All welcome!
Are you working on knowledge extraction from domain-specific sources? Tackling the challenges of long-tail knowledge?
Combining #LLMs and #KnowledgeGraphs to push the boundaries of your research?
1st Workshop on X-TAIL: eXtraction and eXploitation of long-TAIL Knowledge with LLMs and KGs,
co-located with #EKAW2024
Abstract Deadline: September 8th, 2024
https://www.xtail-workshop.org/
@ekawconference #rag #generativeai #explainableAI #longtail #cfp
Can anyone suggest a good (as in accessible for non-specialist readers) article on calibrated trust and AI systems? It's for our @BL_DigiSchol reading group, where papers with equations can be offputting
What do you wish AI, machine learning, or LLMs could do? What's missing from the current picture? What do existing systems get wrong, or fail to do at all? I'm trying to put together a wish list, for a startup oriented towards explainability, responsibility, and a focus on centering human society and human values.
#AI
#ML
#LLMs
#ExplainableAI
#Bias
#Humanism
#Startup
#ArtificialIntelligence
#MachineLearning
#NaturalLanguage
#ResponsibleAI
I actually had someone reach out to me today about starting a company around some of my previous work on "explainable AI" -- machine learning that results in human-interpretable results, as opposed to the inscrutable tables of numbers that make up neural networks. This is pretty exciting stuff to me, because I've wanted to start a business along these lines for a long time. I also got to introduce someone to the concept of "exit to community", so +1 for social activism, too. And the timing couldn't be better, given the layoff I was just caught by.
5 days till deadline for the Human-centered Explainable AI (#HCXAI) workshop at CHI! Pls repost & help us spread the word
Submission pro tips:
1. Explicitly address >1 question (CfP) on the website.
2. Yes, papers NOT dealing with LLMs are fine.
3. Engage w/ past submissions (build on, don't repeat)
4. Position papers must make a well-justified argument, not just summarize findings