AI Futures
A 6-part event series making sense of where AI is heading and what it could mean for our future. No technical background needed.
What is it for?
Where is artificial intelligence headed, and what does that mean for the next 10 years? AI Futures is a 6-part series that works through that question in the open – current systems, how fast they are improving, what could go wrong, and who decides.
Each session pairs a presentation with open discussion. The goal is to leave with an informed view of where AI could be heading and the different futures that seem possible.
Who is it for?
Anyone curious about where AI is going. Students, engineers, researchers, policy people, and professionals who read the main AI headlines and want to get a better understanding of the topic.
In about 15 years AI has gone from a research curiosity to capable agents that write code, pass exams and run tasks on their own. We cover how that happened, why making models bigger kept working better than making them cleverer, and what today's systems still get wrong. Then we look ahead: what the current trends – agents, reasoning, automation of research itself – suggest for the next few years, and a tool for asking how big a deal this really is.
Ask when AI will change everything and you get answers from 2 years to never. This session shows where those numbers come from: who is asked, what they are actually predicting, and which methods have a track record. We look at the mechanism behind the short timelines – AI speeding up AI research – and at where forecasts go systematically wrong. You vote before and after, and leave able to judge a forecast yourself rather than picking the one you like.
Will AI take your job, or just change it? We start with your own profession and work it through, then test the 3 things people say to reassure you – trade, new jobs, human advantages – against what the labour-market and adoption data from 2025–26 actually show. The last part asks what AI does to the economy as a whole, and who captures the value.
Different ways this could go badly: a system that pursues the wrong goal, people using it to do harm, a handful of actors ending up with far too much power, and the quieter one where humans slowly stop being the ones who decide. 10 years ago this was theory; today there are documented cases of models deceiving their evaluators or resisting shutdown in tests, and we look at those. Then the other half – what is being built and tried to prevent each, from a live look inside a model to see what it is actually doing, to the tests and safeguards labs run before release.
Governments, companies and researchers all have plans. We go through the main ones – rules, international agreements, controlling who gets the chips, and whether anyone could actually stop development – and ask what each one assumes, what it covers, and how fast it could work. Then the approaches that don't fit in either box: better institutions, defensive technology, better forecasting.
The “so what for me” session. We recap the timelines and trajectories, then turn to your own life: how jobs are changing and which skills hold up, why working on making AI go well is tractable and neglected, and what the paths look like – technical safety, policy, and everything around them. We look at who is hiring in Switzerland, how to upskill, and how to stay sane while paying attention to all of this. You leave with one concrete next step written down.
Join the next series.
Sign up on Luma, or get in touch if you have questions first.
Stay in the loop.
Follow our events on Luma or join our WhatsApp group to become part of the discussion.