Dr Paul Whaley - Raising the ceiling on AI performance in environmental research - 2nd December 11am, Fylde LT2

Event Details

Raising the ceiling on AI performance in environmental research: the most entertaining presentation on data standards you will ever see, or your money back

There is a lot of optimistic talk about how many research areas are becoming “big data” disciplines. However, the extent to which many disciplines are truly big data enterprises is questionable. From the perspective of AI performance, if data cannot be big, it follows that data had better be good.

The problem we have in environmental research is that, too often, the data is not good. Anyone who has bemoaned a paper for lacking core information about methods, applying the wrong approaches and being published anyway, or using the wrong terms for describing what it did and found, will be aware of the problems we have with data quality. Just like people, AIs need complete, accurate information about how studies were done if they are to be effective at using research.

To solve this problem of data quality, we need to do a lot more formal work on data standards. Well-defined standards encompassing the objects and methods of research are becoming increasingly essential. They ensure a computable record of what people are doing and what they think they have found, which allows the results of scientific work to be evaluated for credibility.

This presentation will take the audience from an informal understanding of data standards to appreciating what a truly comprehensive approach to improving data quality in environmental science would look like, how much work is involved, and who should be responsible for doing it. This work is essential if AI is to fulfil its potential.

Teams connection here