AI Skills Dashboard- Explore Interactive Data
Back in 2016, Geoffrey Hinton, considered by many the God of AI, predicted that radiologists would be fully replaced by AI. He recommended that programs should not train radiologists skills because such skills would become obsolete or unimportant with the introduction of deep learning. Almost a decade later, here we are and radiologists did not go away. In fact, AI augments their roles by reducing the time required for image analysis, enabling them to focus on higher-value tasks. But does this mean that such models fully replaced their domain specific skill? In other words, should they be upskilling or deskilling in radiologist skills? More broadly speaking, the real question then becomes how should workers re-skill, should they be studying AI & LLM’s, computer science theory, or a practical craft like plumbing?
Our goal in this article is to identify how AI exposure influences the importance of skills. Specifically, we are interested in the relationship between our AI exposure measure and changes in skill importance. This will be our so-called AI Upskilling effect. So, what do we observe? We find that there is an increase in skill importance towards human centric and computer science skills. We believe this has to do with the way humans make decisions. From literature, we observe that humans are more cautious about trusting AI outputs especially when decisions stake is high. This is naturally due to the blackbox nature of AI systems and what the actual role of AI outputs are (they are inputs for downstream decision making tasks).
If we go back to our radiologist example, they rely on AI’s diagnostic predictions to make final recommendations. This reliance naturally guides them down different decision paths, which will ultimately influence clinical outcomes. In this dashboard, we can identify AI upskilling and deskilling effects across a wide range of skill subsets and educational requirements. This tool provides practitioners and academics with a means to systematically identify shifts in skill requirements driven by AI exposure.