Recently, I had the great pleasure of participating in the Bureau of Meteorology’s strategy day, where we explored the transformative potential of Generative AI. Reflecting on CSIRO’s experience with digital science transformation and examining our industrial insights into how generative AI diverges from its predecessors was a truly fascinating exchange.

I clarified common misconceptions and highlighted the distinctive features of generative AI compared to other technologies. Some points were:

1️⃣ General Capability: We might need to perceive Language Learning Models (LLMs) not simply as tools or functionalities, but as akin to smart interns.

2️⃣ Ease of Access: Owing to improved accessibility, we are observing a shift towards low-cost, experiment-driven approaches as opposed to traditional top-down, problem-driven/cost-benefit driven ones.

3️⃣ Blurring/Moving Division Boundaries: The “reverse Conway’s Law” – LLMs could potentially alter organizational structures and boundaries, enabling better integration and cross-discipline collaboration.