In the first video edition of The Sydney Dialogue Summit Sessions, David Wroe sits down with Connor Leahy, co-founder and CEO of Conjecture AI. David and Connor speak about the catastrophic risks that a powerful but uncontrolled and unaligned artificial superintelligence could pose to humanity, and Conjecture’s approach to safe AI called “cognitive emulation”. They also discuss what it means for an intelligent agent to have goals, and the likelihood that the current dominant AI approach of large language models can continue to be scaled up with more computing power.

Connor was a panellist at ASPI’s Sydney Dialogue cyber and tech conference held on September 2 and 3. This is the first of a series of podcasts filmed on the sidelines of the conference, which will be released in the coming weeks.

Check out ASPI’s YouTube channel here to watch the full video.