A brave new world awaits, powered by the impressive capabilities of generative artificial intelligence. The idea of clever software doing all your work for you could be equal parts appealing, fascinating, and even frightening. But just what are these jobs that AI is – or will be – taking over?

A simple question which leads into a cautionary note on reaching too high, too far, too soon. Because here’s the rub: while the promise is certainly there, the reality of AI in the workplace today is generally quite different. Particularly when you’re talking enterprise AI.

We’ll talk to an example soon, but first, consider that right now, AI is riding (very) high on hype. There’s no better representation of technology development we can think of than Gartner’s celebrated Hype Cycle, and late in 2023, the market watcher put generative AI right at the pinnacle of the aptly named ‘Peak of Inflated Expectations’.

For those unfamiliar with the market watcher’s Hype Cycle, it is a rather exciting but scary representation of the ‘latest greatest, cutting edge, next best thing’ in the tech space (like cloud, big data, Internet of Things, blockchain, augmented reality, quantum computing, etc.)

Exciting because each phase describes in arguably perfect terms the general perception of these emerging technologies. Scary, because get caught at the wrong moment, and the hype cycle could just as easily represent millions of wasted dollars.

Going into some detail, the Hype Cycle kicks off with the ‘Innovation Trigger’ (chatter about new possibilities from some or another breakthrough). The ‘Peak of Inflated Expectations’ is reached through imaginations running wild, and uninhibited excitement assuming an already-perfected technology.

Only to be followed by a sudden plummet into a dark ‘Trough of Disillusionment’ – because the technology can’t meet those Expectations.

Gartner’s 2024 Hype Cycle released in July suggests that’s where we are heading, and the living proof is – according to Gartner VP Analyst, Mary Mesaglio – nearly half of CIO’s today say AI hasn’t met ROI expectations.

Now, just to take a quick break there. It’s been said the tech industry almost always delivers on its promises, just not in the expected timeframes. Cloud is a perfect example of this. Conceptualised way back in the 1940’s, it was only when the likes of Salesforce, Amazon, and Microsoft launched their cloud divisions that organisations started to transition – and still it’s been a slow road.

This is where the Hype Cycle’s ‘Slope of Enlightenment’ kicks in. A far more gradual process, the Slope represents the industry figuring out how to put this fancy new thing to work. If you are in IT, but in an industry that isn’t being remastered by AI yet, you can afford to go at this more measured pace which should hopefully, lead you to delivering AI outcomes safely and at scale.

For those moving at this pace, the likely focus right now is how to get employee productivity from AI. And this is harder than it looks because to get productivity to increase to levels that equate to ROI takes three things:

1. You have to get people to actually use the new AI tools and keep using them.
2. You also need to match AI outcomes to the right level of job complexity and job experience to ensure that the time saved, or the insights given, are actually beneficial to the user.
3. You need to focus on what Gartner Analyst Mary Mesaglio calls “deep productivity”.

Here’s an example:

Mark handles customer enquiries for his organisation. He’s been in his job for six months and his employer has introduced AI to help Mark and employees in the same role resolve issues faster by augmenting responses to enquiries. This saves Mark around 40 minutes a day. It’s a boost, sure, but as an organisation, how do you know that those 40 minutes saved are actually being allocated elsewhere and not just giving Mark an excuse to slow the pace or take more coffee breaks? Gartner calls this ‘Productivity leakage’ and companies should expect to experience 10-30% leakage when it comes to time savings.

Now consider Janet. Janet has been with the same company for 5 years and does the same role as Mark. She already knows the correct ways to respond to enquiries because she’s experienced. So, for Janet, the AI in this use case isn’t very helpful. She gets very little productivity gains.

Next consider a lawyer. Their work is complex, and AI isn’t always going to give the right responses because understanding law requires deep experience. So, giving Gen AI to a new lawyer who can’t tell a good AI output from a bad one it isn’t going to help them much. However, give Gen AI to an experienced lawyer, who knows what good looks like, and they’ll likely reap significant productivity gains.

This is ‘deep productivity’, and it’s what Gartner notes as the “sweet spot” – the key to unlocking real ROI.

The take-away

Back to our Hype Cycle. Hype doesn’t mean a new technology isn’t promising or lacks value. But as it demonstrates, getting caught up in the excitement can be costly. AI investments are not like conventional IT investments. With Gen AI, it’s really easy to waste money, and just because you can do something using the latest cool tech, doesn’t mean it’s going to bring you value.

The first and most important step to realising value is the business plan and the use case: why are you looking to use AI, and what for? What problem does it solve? Have you asked the people at the frontline if they want or need AI help? What sort of a return on investment will the AI generate? How will you implement and govern it? Simple questions in theory – but considerably complex to unpack and answer.

That’s where NRI can help.

We’ve designed a service that starts with helping you to unpack what AI means for your business. Whether that’s Microsoft Copilot, Now Assist from ServiceNow, Einstein from Salesforce, Joule from SAP, Oracle AI, or others.

Our approach starts with helping you to understand your users so you can determine where value can be found, and then developing a vision of how AI will help you achieve your goals. Then, get ready to be ready with a full technical assessment across things like data quality, security, privacy, and governance. A crucial step that helps bring to the forefront any risks or gaps that could affect performance. All of this, right through to implementation and change management is designed to put you on the path to least resistance. And genuine ROI.

Download our latest guide ‘Make Microsoft Copilot work for you’.

We’ve documented some the key mistakes organisations are making when it comes to rolling out Copilot, and the practical steps leaders can take to turn experimentation into tangible business impact.