> ## Content Index
> Fetch the complete content index at: https://thinking-in-public.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI (conferences) - 40 years on
- URL: https://thinking-in-public.ghost.io/ai-conferences-40-years-on/
- Published: 2026-10-14T13:10:42.000Z
- Updated: 2026-10-14T13:10:42.000Z
- Author: Virginia Dignum

Today, we wrap up [AIES 2026](https://www.aies-conference.com/2026/?ref=thinking-in-public.ghost.io), for which I had the honour of being general co-chair. A great event, with many deep discussions and very active engagd, and seen from my perspective, young participants. 

Which got me thinking of my first AI conference, in November 1986: the second [IFIP WG 2.6 Working Conference on Data Semantics, "Data and Knowledge" (DS-2)](https://dblp.dagstuhl.de/db/conf/ds/ds86.html?ref=thinking-in-public.ghost.io), in Albufeira, Portugal. I was an MSc student working with Amílcar Sernadas, the conference chair, who asked me to help with the practical organisation. It was my first view of a scientific meeting from the inside, and a good lesson in how much of science rests on volunteer work, logistics and goodwill. (Note to self: I've been organising scientific events big and small for 40 years! Stop now!)

The venue was a beach resort in the Algarve, mostly empty besides for the conference participants as it was autumn. Between sessions, and long after them, I talked with [Frank Dignum](https://share.google/Ho9xRckIkfnAWJo5V?ref=thinking-in-public.ghost.io). We spoke about logic, about Africa, about the differences between Portugal and the Netherlands, about the future, about research and about our ambitions. I did not know then how far those conversations would reach but 40 years on, we are still not finished (and are married for 39 of those years)

Somewhere in a box in Amsterdam, we still have 2 copies of the preproceedings: A4 photocopies in a clear plastic cover, held together by a plastic comb binding. Speakers used transparencies on overhead projectors, many of them handwritten. As far as I can tell the full volume is not available online. The dblp entry lists only the contents. A reminder of how much of the earlier AI work is nowhere to find outside library shelves. 

Reading the table of contents today is instructive. The central question was how to bring databases and knowledge representation together, so that systems would have some grasp of what their data mean. Papers addressed deductive and recursive querying, formal specification of information systems, natural language access, and reasoning about knowledge, communication and time. 

Much of this has returned in new form, or actually was never ever gone. Knowledge graphs and ontologies revisit the boundary between data and knowledge. Datalog-style reasoning is back in practice. Epistemic and temporal logics sit at the core of multi-agent systems. Text-to-SQL with large language models was 1980s version of the complexity of natural language access to data, with the same lesson: fluent output does not guarantee a faithful formal semantics.

At AIES 2026, the discussions in the room are not very different from those in Albufeira: what do systems know, and what do their data mean? What has changed is the question we did not ask then, at least not explicitly: who is accountable for what these systems conclude and do? Accountability, provenance and norms have turned semantics from a technical problem into a governance problem. We have also moved away from the old assumption that meaning can be fully specified in advance by a closed logical model has given way to learned representations, and are left with the challenge from specification to verification and alignment. That shift is, in many ways, what my work on responsible and governance-aware AI addresses, and is still long not finished. 

---