After soaring levels of investment, there is real pressure mounting on technology leaders to make AI work. However, as the Harvey Nash Digital Leadership Report and our latest Tech Flix documentary ‘The AI Paradox: AI is Scaling, Skills are Not’ explore, the vast majority of organisations are overlooking a vital success factor in not upskilling people in AI skills. So, as a digital leader, what are the conversations you should be having with your peers to better harness the huge potential in these tools?
2025 was meant to be the year organisations unlocked real return on investment, but data from our Digital Leadership Report pointed to just 33% of organisations reporting “demonstrable ROI beyond initial piloting”.
That lack of ROI isn’t dampening enthusiasm. In fact, quite the reverse, as the demand for AI skills jumped by 82% year-on-year, the biggest jump in any skills scarcity we’ve ever seen.
But what do we mean by AI skills?
We know vibe coding is a growing phenomenon that is widening technology’s already broad church, blurring the lines of what skills we’d consider part of the DNA of an engineer. Frankly, who doesn’t use AI in their day-to-day work in some capacity?
Pressure to deliver and soaring expectations have led to technology and digital leaders finding themselves crippled by indecision. The landscape is more complex, more nuanced, and changing quicker than ever.
So in this climate, what do digital leaders need to talk about?
1. The conversation when managing-up: "Slow down and get specific.”
Fear of Missing Out (FOMO) is placing immense pressure on digital leaders, and simply telling them the strategy should be to “do AI” isn’t helping. As basic as that sounds, it’s the lived experience of many. The challenge for the leader is to push back against vague adoption and demand that there is a focus on specific business pinch-points.
In our Tech Flix documentary, Phil White, MD of Audacia, argues that organisations should actually slow down rather than speed up. He warns that FOMO will leave too many getting “on this AI bandwagon without actually realising what it is, what it can achieve, what it can do”.
Josh Nesbitt, CTO at Genio, lends weight to this rationale by adding that AI is too broad a term to actually be useful in conversation. Instead of talking about AI generally, leaders need to start talking about it in relation to specific problem-solving.
Laura Gilbert, the Senior Director of AI at the Tony Blair Institute, expands on the challenge with a vital additional consideration. Because there is so much pressure to move now, and because we are terrible at predicting the future, leaders run the risk of building inflexible systems based on current hype.
Laura argues the conversation should centre on flexible systems rather than betting on a specific outcome, giving you much more room to adapt as an organisation.
2. The conversation with fellow executives: “Have you tried AI for yourself?”
When AI succeeds or fails, it’s the digital leader who is held accountable. But what’s been clear for some time now is that they are far from being the only accountable figure when it comes to the implementation of new tools. That’s problematic because non-technical executive colleagues lack the proper understanding of what’s technically realistic, leading to poor decision-making.
There is also the practicality of understanding what you’re being sold. Every vendor will have baked AI into their pitch, and with FOMO so dominant, it’s important to understand if you’re being sold something useful, or as Laura Gilbert puts it in our latest Tech Flix documentary, “entirely wrong at a terrible price”.
Laura argues that things go wrong when leadership doesn’t use the tech themselves. Her advice is to ambush colleagues and “shove a phone in their hand”, telling them that you’ve made them an account and to give it a go themselves.
For Josh Nesbitt, it’s not just a technical issue, but a cultural one. Similarly to Laura, he feels leadership must be hands-on, and “model usage”, so that the wider organisation also feel there is permission to experiment and fail.
3. The conversation with your L&D Manager: “Most people know how to start the engine, but do you know where to apply the brakes?”
As our recent documentary explores, there is real confusion over what we actually mean by AI skills and what upskilling involves. An engineering function might have one perspective on AI and its function, but AI means anybody can now build tools and products regardless of their ability to code. Professionals across an organisation (sitting outside the technology) can learn prompt skills that unlock opportunity, but with little awareness of the potential risks.
Nicky Danino, Leeds Trinity University’s Head of School of Computing and Creative Industries, uses a Formula One analogy. Whilst the vast majority of us can be AI ‘drivers’, you will still need a small team of engineers who can “get under the hood”. Our conversation with Philip White, Managing Director at Audacia, helps to further advance what a driver really does. It’s about understanding the application of AI. Drivers need to understand what AI is good at (forecasting and analysing data, etc.), and more importantly, what it’s bad at. That’s far more valuable to an organisation than theoretical knowledge.
I’ve also been told that the government hopes to give foundational knowledge to millions across the United Kingdom with courses lasting two and a half hours. Will Abbey, Chief Commercial Officer at Arm, challenges that idea. To unlock true productivity, you can’t gloss over basics like data manipulation, even if you’re not actually writing code yourself.
What we’re saying is that making the car go (like driving) is only half the challenge. A good driver also needs good road awareness, and that experience cannot be faked. Without it, accidents are highly likely.
4. The Conversation with HR: “We might be efficient today, but we’re building a capability crisis for tomorrow.”
The drive to find efficiency has pushed organisations to automate a range of traditionally entry-level tasks such as coding, copywriting, and admin. It’s widely reported that in sectors like professional and financial services and legal, the number of junior roles is reducing. In the short term, organisations only feel economic upside, but the medium to long-term might look very different as they intentionally break their own talent pipeline.
Using technology as an example, Nicky Danino stresses the fact that if AI does take over entry-level programming, “there is nobody to then check that AI is doing it properly”. A human in the loop is the handbrake we widely talk about as being critical in the implementation and adoption of AI. But as fewer people hold the necessary skills to query AI’s work, that will naturally get harder.
Phil White also points to the structure of an organisation shifting, moving from a triangle to a diamond-shaped organisation, where there are increasingly disproportionate numbers of staff in leadership roles, and a lack of mid-to-senior talent being developed.
Over a Genio, Josh Nesbitt is using these trends to their advantage. He’s not cutting back on the number of junior hires, but instead focusing on using AI to augment them to do more. In his opinion, the conversation has to be about output, not reducing headcount.
Whilst the temptation might be to scale back, the long-term competitive advantage will probably be gained by leaders brave enough to move in the opposite direction.
5. The Conversation with your people: “We want to remove friction, not meaning and purpose.”
Finally, no surprises, we have to address the fear among staff that AI is there to take jobs, which in turn makes them suspicious and hinders adoption. The common refrain is AI won’t take your job; the person sitting next to you who uses AI will take your job. But even that isn’t helpful if we’re looking to build positive sentiment towards these tools. We want to foster collaboration, not competition.
Will Abbey told us that at Arm, AI is being used to review documentation and do debugging tasks, the sort of routine work that can take up to 30 or 40% of an engineer’s time. That time can be given back to design and working on new ideas and solutions, essential to an engineer’s overall job satisfaction, given it’s inherently creative. In their sales teams, AI can be used to generate quotes for customers. That means Arm’s staff spend more quality time actually engaging with their customers, and are again more likely to come up with ideas that can benefit Arm and the customer.
One note of caution. It’s often assumed that AI simply removes monotony, but that isn’t always the case. If part of someone’s role is creating a document, and they find that process engaging and creatively satisfying, replacing it with editing AI-generated drafts may not feel like an upgrade.
Technology leaders need to take a more nuanced view of what actually motivates their people. Not every manual task is meaningless, and not every automated task increases job satisfaction. The impact of AI on engagement will depend on how thoughtfully it’s implemented, not just how efficiently it performs.
What does success look like after you’ve positioned AI with your colleagues and stakeholders?
The underlying themes of our Tech Flix documentary point to a few key principles. Rather than being a transactional leader buying tools, you are instrumental in helping shape the culture of your organisation, far beyond the boundaries of a traditional CIO or CTO.
I started this article talking about crippling indecision. What I hope is that through our conversations, we’re able to point to examples of clear thinking that can help leaders move debate to action.
If you stop vague discussions about ‘AI skills’ taking place, start talking about problem solving, if you can get your peers to actually engage with the tools themselves, and protect your talent pipeline, then, rather than being caught in a spin by the complex environment we’re all operating in, you can remain a competitive and innovative organisation.
If you’re serious about moving from AI ambition to AI outcomes, it starts with talent. Explore how we help find the best AI talent and support your organisation in building high-performing AI and ML teams.
