Alex Clarke
I'm currently a software developer at SKAO, working on the data management stack for the SKA Regional Centres (SRCnet). This will manage the processing, collection and distribution of SKA data to astronomers across the world. Before that I did a PhD using LOFAR to observe radio galaxies and merging galaxy clusters, and then a post-doc looking at applying machine learning models to source classification problems in multi-wavelength surveys. You can learn more about me at informationcake.com.
Session
As implementing complex AI models with massive datasets becomes increasingly accessible, at what point do we reach the limits of explainability? "Simple" problems are easily tackled by AI when input parameters are well-defined and the underlying processes are understood. However, as datasets grow and the underlying processes become exponentially ill-defined, we must question whether we are too quick to abandon explainability in favor of rapid results. While societal pressures often encourage this trade-off, the scientific community must resist it. Furthermore, our human drive to seek out rules that lead to definitive conclusions can be misleading. Studies have shown that rats can easily outperform humans at certain pattern recognition tasks, illustrating how our own cognitive biases can interfere with objective analysis.
This presentation will explore these challenges using examples analysing over 100 million astronomical sources with images, spectra and metadata via multi-modal AI models. I'll show how an abundance of information can sometimes yield increasingly misleading results, the importance of a multi algorithm approach, and how developing user interfaces packaged with your research is becoming a valuable tool in distilling trust and value to the consumers of your publications. Ultimately, mitigating bias and presenting trustworthy, reproducible and transparent results are key responsibilities of human problem-solving in the age of Big Data and AI.
Check out my interactive results here: sdssexplorer.informationcake.com