Speaker: Paul Bellette
Affiliation: University of Queensland
Abstract
Large language models are making it dramatically easier to produce software. Things that once required hours or days of programming can increasingly be built in minutes, often by people with relatively little conventional software experience.
This is kind of amazing, but will inevitably have some fallout.
I’ll talk about this shift from the perspective of an applied mathematician and engineer who has spent much of his career building software for messy real-world systems, and who is now building a company in the middle of the current AI transition. I’ll show some examples of what modern models can do and where they are still surprisingly unreliable. I also want to talk about why I think this is potentially a very good time to be a mathematician.
About Industry seminar series
The industry seminar series connects industry and academia.
Speakers from industry present challenges and opportunities in their work, along with their insights.
To receive news on the seminar series, subscribe to the mailing list (UQ email address required) or send an empty email to maths-industry-seminar-join@lists.science.uq.edu.au