How to talk to your board about AI without overpromising
Most board AI conversations overpromise and underdeliver. Here's how to have a more honest conversation about AI governance, risk appetite, and strategy.
There’s a conversation happening in boardrooms across Australia right now, and most of them are going roughly the same way. The board wants to know what the company is doing on AI. The CTO or delivery lead has to answer that question honestly. And the honest answer is… complicated.
Not because the technology isn’t real – it is, and the pace of change is genuinely significant. But the gap between what AI can do in a demo and what it can do reliably, at scale, inside your actual data environment is much larger than most board decks let on.
That’s where things tend to go sideways.
The temptation is understandable. Boards are anxious about being left behind – they’ve read the headlines, they’ve seen what OpenAI and Google are shipping, and they want reassurance that the business isn’t asleep. So the deck gets optimistic. The pilot gets called a programme. The proof of concept gets positioned as a platform. And twelve months later, when the POC hasn’t made it to production and the ROI hasn’t materialised, everyone’s wondering what happened.
What happened is that the boardroom conversation set the wrong expectations from the start. It’s a pattern we see consistently in enterprise Australia’s AI conversations – and it’s avoidable.
What boards are actually asking for
Here’s the thing – boards aren’t asking for hype. What they’re actually asking for is confidence – confidence that the organisation has a clear-eyed view of the opportunity, a realistic plan to get there, and the governance in place to do it safely. Those are three genuinely different things, and each of them needs a different conversation.
Talking about the opportunity: Be specific, not sweeping
“AI will transform our business” tells the board nothing useful. “We’ve identified three use cases where AI can reduce manual processing time by 40%, and here’s how we’re prioritising them” – that’s a conversation worth having. The organisations getting real ROI from AI right now are the ones that picked the right problem first, not the ones who started with the most ambitious technology vision and worked backwards from there.
This is also where the AI implementation gap tends to open up – between what leadership believes is happening and what’s actually in production. Being specific in the boardroom is how you close it.
Talking about the data foundation: Don't skip the hard part
This is the part most board conversations tend to quietly skip over. AI is only as good as the data it runs on, and if your golden metrics are fragmented across systems, if your data pipelines aren’t clean, or if you’re operating in a regulated environment with data sovereignty constraints – that needs to be on the table. Not as a reason to do nothing, but as a realistic view of what “production-ready AI” actually requires and what it would take to get there.
Talking about AI governance: The question every board wants answered
AI governance is the one most boards want to hear about but few delivery leads feel confident addressing. AI risk appetite isn’t abstract – it’s a concrete question: what decisions are we comfortable letting AI make, and where does a human stay in the loop? The organisations that are building real trust with their boards on AI are the ones who can answer that clearly, with guardrails built into the architecture rather than a policy document sitting in a folder somewhere.
At the Data & AI Summit in Sydney earlier this year, this tension came up repeatedly. The organisations presenting the most credible AI programmes weren’t the ones with the most ambitious slide decks. They were the ones who could articulate the failure modes, describe the assurance frameworks, and explain what they’d learned from the POCs that didn’t make it – which, given that around 80% of POCs never reach production, is most of them. It’s consistent with what Australian companies are learning about AI at scale.
That’s what trust actually looks like in a board context. Not certainty. Credibility.
The CTO who walks into the boardroom and says “here’s what we know, here’s what we don’t, and here’s how we’re managing the risk” will have a much more productive conversation than the one who says “AI is going to be transformative for us” and leaves it there.
Four things that actually help
If you’re preparing for this conversation, here’s what tends to make the difference:
Set the AI risk appetite explicitly.
What’s the organisation’s actual tolerance for AI-driven decisions? Document it and make it real – that’s the foundation for everything else, including which use cases get prioritised, what governance looks like, and how you report back to the board.
Separate the pilot from the programme.
Pilots are for learning. Programmes are for scale. Don’t conflate them. The board needs to know which one they’re funding and what the criteria are for moving from one to the other.
Talk about the data foundation honestly.
If it’s not ready to support what you’re proposing, say so – and explain what it would take to get there. Boards can handle a realistic timeline. What they struggle to handle is being surprised eighteen months later when the AI strategy stalls because the data wasn’t ready.
Build guardrails, not guidelines.
A document that says “AI should be used responsibly” is not an AI governance framework. Guardrails built into the architecture – restricting what the model can do, keeping humans in the loop at the right points, creating an audit trail – that’s governance.
The board conversation about AI doesn’t need to be the one where you promise the moon. It can be the one where you demonstrate you’ve actually thought this through. And honestly, that’s a much more interesting conversation to be in.
At Restive, this is the kind of work we do with technology leaders every day – helping organisations move from AI ambition to AI reality, with the strategy, governance and team capability to back it up. If you’re navigating this conversation and want a sounding board, we’d be happy to talk.
Related articles
