GenAI Summit Talk: “Meaningful” Human Oversight for GenAI

🤔 What does “𝒎𝒆𝒂𝒏𝒊𝒏𝒈𝒇𝒖𝒍” really mean in “meaningful human oversight” when it comes to GenAI? How can humans oversee something they don’t fully understand? What are the approaches?

It was a pleasure to explore these questions in my talk earlier today at the GenAI Summit Sydney. I debunked a few common myths—like “human-in-the-loop solves everything,” “AI are just tools, humans make the decisions,” and “AI-augmented intelligence.”

I introduced CSIRO’s Data61 approach to design-time oversight—ensuring human agency and oversight are properly embedded in GenAI—and runtime oversight through guardrails and system-level understanding, rather than just model-level understanding. Find more here https://lnkd.in/gPhid9tX

Here are a few selected slides highlighting the key messages. https://www.dropbox.com/scl/fi/g5gr5ug2o5pbjkkb9f35f/selected-Zhu-Human-Oversight-GenAISummit20230814.pdf?rlkey=7jnc05g0c5i86245345vj2oxt&dl=0


About Me

Research Director, CSIRO’s Data61
Conjoint Professor, CSE UNSW

For other roles, see LinkedIn & Professional activities.

If you’d like to invite me to give a talk, please see here & email liming.zhu@data61.csiro.au

Featured Posts

    Categories