Dr Brian Lowe - Chief AI Officer in Australia

About this episode

Brian Lowe has held the title of Chief AI Officer—one of the few in Australia to do so—and brings a clear message for business leaders concerned they are falling behind on AI: you are not missing out. In fact, up to 95% of AI initiatives never reach production, and much of what is presented as progress is often little more than hype.

Brian began his career with a PhD in Information Systems and has since built deep experience across enterprise AI and data. His roles have included Assistant Director at the Bureau of Meteorology, CTO of an AI startup, Head of Data Architecture at one of Australia’s big four banks, and Director at a major university, where he helped the institution navigate the early emergence of AI. Most recently, as a Chief AI Officer, he advised organisations across government, utilities, and global retail. He is now focused on building a sovereign Australian AI model—trained on Australian data, hosted domestically, and governed by Australian law.

Brian on LinkedIn: https://www.linkedin.com/in/brianslowe

In this episode, Brian explains what a Chief AI Officer actually does day to day, and much of it involves cutting through noise and challenging misplaced optimism. He has seen conferences showcase questionable case studies, consultancies promote solutions that fail to deliver, and executive sponsors persist with struggling initiatives to avoid admitting failure. His approach to navigating this landscape draws on an unexpected source: a lesson from a Simpsons episode about a monorail.

Then he takes apart agentic AI. Much of what is sold as agentic is just automation, a technology that has been around for 20 years, rebadged with a bit of AI sugar on top. The properly agentic systems are useful but they misfire, and his test for where you can trust them is simple. Would you hand your online banking to an assistant who gets it wrong 3% of the time?

He also explains why the US government accidentally handed Australia the business case for its own AI model, why students who learn with AI do worse, why everyone running a team of AI agents has become a manager whether they like it or not, and why AI has made hiring good people harder than it has ever been.

ALSO IN HERE

- Why beer ended up in the nappy aisle, and what it says about your data
- The "bad debtors" his team discovered that you actually want to keep
- The AI he instructed to abuse him, and why it improves his thinking
- Why the best AI implementation he has ever seen won a Nobel Prize
- Why a $70k graduate might beat your AI subscription

CHAPTERS

0:00 Who is Brian Lowe
1:15 He is building an Australian AI model
2:00 The two events that made the case for it
5:33 "95% of an AI project is data"
7:36 Start from scratch, or fork and fine tune
10:15 Most business problems do not need model training
11:27 The Industrial Revolution question
14:57 A four day week, if we get this right
16:45 What a Chief AI Officer actually does
18:50 "There's a lot of bullshit out there"
23:10 How to spot snake oil before you pay for it
25:22 The monorail episode
28:01 95% of AI projects never make it
30:18 "You are not missing out"
33:52 Rebadged RPA with AI sugar on top
35:03 Beer in the nappy aisle
40:09 What agentic AI really is
43:31 The online banking test
45:00 The best AI implementation he has ever seen
49:18 The infinitely patient tutor
53:06 Students who learn with AI do worse
55:46 The AI he built to abuse him
56:52 Everyone is becoming a manager
1:02:08 Why teams cap out at seven
1:04:54 Why AI made hiring harder
1:09:44 The $70k graduate vs the API bill
1:15:08 How AI hacked your brain
1:17:44 "You cannot substitute people for AIs"