IA on AI - The Internal Audit Future (or not) - The Audit Podcast

Summary4 min read

Podcast Summary: The Audit Podcast – “IA on AI: The Internal Audit Future (or not)”
Host: Trent Russell
Date: July 22, 2026

Episode Overview

This episode of The Audit Podcast delves into the implications and future trajectory of artificial intelligence in the internal audit profession. Host Trent Russell reflects on a LinkedIn post by Tom McLeod—a recognized thought leader in AI and internal audit—highlighting Tom’s bold predictions and healthy skepticism about the role and redesign of internal audit amid accelerating technological change. The episode encourages auditors to critically examine both the opportunities and limitations of AI, and to foster a culture of visible humility rather than ungrounded certainty.

Key Discussion Points & Insights

1. The Largest Redesign in Internal Audit History

Main Point: Tom McLeod posits that internal audit is on the verge of its most significant transformation, largely driven by AI advancements.
Quote (Tom McLeod, via LinkedIn):

"Internal audit is approaching the largest redesign in its history."
[02:10]
Trent agrees, acknowledging that while this is a widely-shared view, the speed of change means some predictions will inevitably prove wrong.

2. Full Population Testing vs. Strategic Sampling

Main Point: Tom predicts that “full population testing” will render sampling a matter of choice, not necessity, owing to AI’s capacity to analyze all transactions.
Caveat: This benefit is contingent; examining everything is futile if auditors fundamentally misunderstand the transaction, its context, or underlying risks.
Discussion: Trent underscores the need for management buy-in and strategic decision-making:
- “Continuous auditing, full population testing has been around, at least conceptually, forever. … I just don't know if management wants full population testing for seemingly every control.” [03:15]
- Empathizing with management, he likens excessive full-population testing to perpetual third-party audits, which would quickly generate fatigue and resistance.

3. The Promise and Peril of Continuous Assurance

Main Point: AI-enabled continuous assurance should, in theory, deliver early warnings of emerging risks.
Tom’s Warning: Such systems may become “a vast organizational nervous system that detects everything, understands little and makes everyone responsible for nothing.” [05:10]
Analysis: Trent notes the potential for overwhelming management with noise rather than clarity.

4. The Death of the Audit Report?

Main Point: Tom anticipates the formal audit report will lose centrality as real-time dashboards become standard.
Counterpoint: There will always be a need for a decisive moment, a “named human being” to declare, “this is what I believe.”
Discussion: Human judgment remains essential:
- “Just because you build a dashboard, that doesn't mean anybody's going to look at it. … There's dashboards with 13 pages on it and somebody looks at one graph on one of those.” [06:50]

5. Tom McLeod’s AI Promises Framework

Structure: Tom presents seven AI promises, each with corresponding reality checks, spanning:
- Testing
- Planning
- Timing
- Reporting
- Talent
- Relevance
- The Board

Trent urges listeners to review the full post for detailed insights and community comments.

6. The New Mandate: Visible, Useful Humility

Quote (Tom McLeod, LinkedIn):

“The profession does not need another person selling certainty about AI. It needs people prepared to place their most valuable convictions where they can be tested, damaged, and when necessary, destroyed.” [08:20]
Tom’s Approach:

“If I am right, internal audit will be rebuilt. If I am wrong, I want to be wrong early, visibly and useful.” [08:30]
Guiding Principle:

“Getting the future wrong is forgivable. Watching the evidence change and continuing to sell the old conviction because your reputation depends on it is not.” [08:50]

Notable Quotes & Memorable Moments

Host’s Perspective on Management Buy-in:
“Imagine if literally every audit we ever did and every procedure within every audit we ever did was audited by a third party … I just don't know if management wants full population testing for seemingly every control.” [04:00]
On Dashboard Utilization:
“Just because you build a dashboard, that doesn't mean anybody's going to look at it.” [06:55]
On Humility in the Face of Change (Tom McLeod):
“The profession does not need another person selling certainty about AI.” [08:20]
“Getting the future wrong is forgivable. Watching the evidence change and continuing to sell the old conviction because your reputation depends on it is not.” [08:50]

Timestamps for Key Segments

00:00: Introduction and context; Tom McLeod’s influence
02:10: Internal audit’s impending redesign
03:15: Full population testing—promise and pitfalls
05:10: Continuous assurance and management overload
06:50: Dashboard adoption and the persistence of human reporting
08:20: Tom McLeod on humility and intellectual honesty in audit
08:50: Closing reflections on change and conviction

Conclusion

The episode encourages auditors to look beyond the hype of AI, questioning not just what is possible but what is practical, sustainable, and valuable. Trent and Tom champion a forward-thinking, yet humble stance—one that expects and embraces being wrong, as long as corrections are made swiftly and transparently. The conversation underscores the ongoing need for critical human judgment and the importance of clearly communicated, well-founded convictions in a rapidly changing profession.

For further insights: Explore Tom McLeod’s original LinkedIn post for a deeper dive into each AI promise versus reality.

Loading summary

Transcript1 lines

[00:00]
A
Welcome to the IA on AI Podcast, part of the Audit Podcast Network, where we bring you weekly updates on AI from the internal auditor's perspective. Hey, everybody. This week, instead of going over some AI related news, I wanted to point you to A post on LinkedIn from Tom McLeod. Tom has been on the Audit podcast. He's been on other podcasts as well. So if you search Tom McLeod, that's McLeod. I'm sure you can find plenty from Tom. He also posts, it seems like every day on AI and internal audit. And so a really good follow if nothing else, and especially that's kind of what we're going to do here in a minute, is just to get people thinking about the future of internal audit with AI. So I'm not going to read the entirety of the post and if you are watching us on YouTube, then you can see the post that I'm talking about. So I'm just going to read the first two, three sentences and then we'll kind of dig into it. Tom says I may be wrong. Internal audit is approaching the largest redesign in its history. Personally, I think we're all probably agreeing to that. And let's be honest, many of my strongest views about that redesign may be wrong. And with as fast as things are changing and the number of points, I think there's like 10, 12 points that Tom makes, some of them are going to be wrong. I just wanted to pull out a few of those and like I said, if nothing else, get people thinking about what the future is really going to be like. Tom also used NotebookLM to create this nice visual. So again, if you were watching my screen, we'll blow it up for everybody. All right, so one of Tom's first statements, he says, my belief, full population testing, sampling is becoming a choice rather than a constraint. We can examine every transaction. And so Tom puts like his belief what he thinks is going to happen, and then he also states why it may be wrong. And so he goes on to say why it may be wrong. Examining every transaction does not help if we have fundamentally misunderstood the transaction, the population, or the true nature of the harm or the risk. Totally agree, and I think Tom has said something similar. But continuous auditing, full population testing has been around, at least conceptually forever. And I know in doing continuous auditing, continuous monitoring projects with clients that whenever the solution is built and then it goes to management, they go, awesome. We are not going to sit here and get audited for every transaction. And then you send us a sample of 25 every month, quarter, whatever it is, it Just doesn't work out very well without management's buy in. And so the way I kind of illustrate this so that we can empathize with management. Imagine if literally every audit we ever did and every procedure within every audit we ever did was audited by a third party and then just constantly them going, this is wrong, this is wrong, this is wrong. You got to do this, you got to fix this. I just don't know if management wants full population testing for seemingly every control. So I think we do need to be strategic in where we do full population testing going forward. One of Tom's other beliefs, continuous assurance, will seamlessly produce earlier warnings for emerging risks. I think that makes sense, that's logical why that may be wrong. Tom goes on to say, it could create a vast organizational nervous system that detects everything, understands little and makes everyone responsible for nothing. I could definitely see this being played out like that. Also similar to the if we are doing continuous assurance or we're testing full populations, there's going to be so much much that what, what is management and what are we really going to pay attention to? Tom's other belief is the death of the report or the audit report. The formal audit report will lose its central place as real time dashboards take over. Why Tom may be wrong. Organizations still need a moment when analysis stops and a named human being is required to definitely say, this is what I believe. I agree with this 100%. I think there's still going to have to be people having the conversations. So agree with that part. The other thing in building dashboards for management and internal audit and talking to teams who have built hundreds, probably thousands at this point. I promise you, just because you build a dashboard, that doesn't mean anybody's going to look at it. One of the number one questions we ask when people start talking about dashboards that they've already developed either for internal audit or management, I go, can you see who's using that dashboard and how often? And most of the time you can see at least what's being used and it is very little. I mean there's dashboards with 13 pages on it and somebody looks at one graph on one of those. But dashboard design aside, I still agree. I think there's going to have to be the human in there. There's seven of these AI promises and the reality of that. So what Tom thinks and then where he could be wrong, like I mentioned, so there's seven of those. Go check out the post and read more in there. There's always really good stuff in the comments of those Also, if you look at it though, Tom does these across testing, planning, timing, reporting, talent, relevance and the board. So it's pretty interesting in the way that he came up with the structure of that. And then because this close is better than anything I could come up with, I'm going to quote exactly from Tom's last slide from this LinkedIn post. He says the new mandate Visible, useful humility. The profession does not need another person selling certainty about AI. It needs people prepared to place their most valuable convictions where they can be tested, damaged and when necessary, destroyed. If I this is Tom here. If I am right, internal audit will be rebuilt. If I am wrong, I want to be wrong early, visibly and useful. And then I really like the way Tom wraps up this whole thing. So to close us out, Tom says getting the future wrong is forgivable. Watching the evidence change and continuing to sell the old conviction because your reputation depends on it is not. Thank you for listening and be sure to follow the link to greenskiesanalytics.com in the show notes and schedule time to see how green skies can make the hype of AI a reality in your internal audit department.