Artificial intelligence is reshaping the very concepts of science and education: the world of elegant equations is giving way to patterns extracted from data and ever more accurate predictions, and the question “why?" is increasingly replaced by “what will happen?". AI excels at interpolation but is even weaker at extrapolation, so the future will likely be hybrid — physics-informed neural networks together with classical insights reverse-engineered from model outputs may form a dual methodology. Engineering seems to be drifting towards medicine: safety factors and heuristics are being displaced by clinical-test-style validation, digital twins, and evidence-based engineering. New legal questions are also coming to the fore — from the copyright status of training data to the personality-rights implications of voice cloning — and the scientist's role is being rearranged: the classic “interpreter" / “explainer" is being replaced by the “toolmaker", and it is an open question whether the explanatory role returns in the future. The transformation of the labour market will also be faster and deeper than any previous technological shift: European social systems were not designed for a persistent 20–40% structural unemployment, so rethinking the link between work and income may well land on the agenda. The talk will cover these and many other topics — time and venue:
27 November 2025, 16:00–18:00, BME Department of Computer Science building, IB210, 1117 Budapest, Magyar Tudósok körútja 2.