Thanks for the invitation to give a talk. I guess we should pick a topic you'd be interested in for the talk so I can prepare the title and abstract.
These are the topics we've been working on and thinking about in recent years:
should there be an uncertainty penalty in expert elicitiation of rare-event probabilities,
combating indubiety and the crisis of trust in the age of fake news and flat earth conspiracists,
how to use confidence calculations in designing smart regulation,
imprecision statistics: quantitative methods for the next century of big data
optimizing distributed decision making and empowering the developing world to avoid roadway carnage,
computing with confidence via confidence boxes (going directly from random sample data to p-boxes),
methods to protect privacy yet allow research on sensitive data,
sensitivity analysis of probabilistic risk assessments (it's way trickier than you thought),
shape analysis (uncertainty analysis in forensics),
detecting clusters among very rare events,
how risk analysis matured in regulation of electricity generation and distribution,
risk communication (expressing uncertainty so people can natively understand it),
decoding uncertainty in natural-language expressions,
incorporating uncertainty in early spacecraft design at NASA using p-boxes,
ecological/environmental risk analysis applications of p-boxes,
validation of scientific/engineering models and advanced bias correction.
Any of these topics seem interesting or useful?
Titles and abstracts for recent talks are archived at https://sites.google.com/site/scottfersontalks/. It's pretty easy to reprise anything from this list that would be appropriate.
I'd certainly be interested in any general hints you might have about the number of minutes I have, the size of the audience, the technical background/interests of the audience, any prejudices they might have, third-rail issues to avoid, etc. Any other advice or suggestions?