CIOs urged to pace AI adoption

Artificial intelligence (AI) dominated the agenda at the recent Gartner IT Symposium 2024 in Gold Coast, with industry experts calling for CIOs to pace their adoption of the technology as they navigate the fast-moving AI landscape.

Mary Mesaglio, Gartner distinguished vice-president analyst, emphasised that industries are experiencing varying levels of disruption from AI. While an “AI-steady” approach may suffice for organisations focused on productivity gains, those seeking to transform business models or become AI leaders should embrace “AI-accelerated” strategies.

One example of an AI-steady adoption can be seen in the Victorian Government’s use of the Archistar software to check the compliance of building permit applications, said Gartner distinguished vice-president analyst Kristian Steenstrup. Compliant applications and prompt processing with deep productivity mean happy citizens and happy builders.

On the other hand, the pharmaceutical industry is more open to disruption from the introduction of AI, with the potential for faster product introductions, new revenue, improved asset yield, and greater productivity.

Steenstrup gave the example of pharmaceutical giant Eli Lilly, which is looking to achieve the equivalent of 2.4 million digital worker hours in 2024 alone. Developing new products more quickly would result in increased revenue, while improved productivity reduces costs.

One of Gartner’s top strategic trends for 2025, agentic AI, marks a significant shift. Instead of advising or assisting a human, an AI agent acts on behalf of the organisation. It might create and distribute a request for proposal, evaluate the responses, and then place a purchase order. “This is a very big shift,” observed Mesaglio.

However, organisations need to be aware of the limitations of agentic AI when it is added to the software they use, and they should avoid allowing it to become the next generation of shadow IT or falling into the trap of thinking it is just a form of robotic process automation, she added.

Despite the enthusiasm for AI tools, integrating them effectively remains a challenge. According to Gartner’s research, 98% of IT leaders said employees are eager to try new AI tools, but 72% said those employees struggle to integrate AI into their daily work. Additionally, concerns linger over “productivity leakage” from employees taking additional or longer coffee breaks.

Also, the gains from the adoption of generative AI (GenAI) vary between employees and roles, observed Steenstrup. It is best applied to situations where job complexity and experience are either both low or both high, because that is where deep productivity can be achieved. Relatively small benefits accrue when the technology is applied to experienced workers in low-complexity jobs, or to inexperienced workers in complex jobs.

This means behavioural issues must also be taken into consideration. If an employee with six months’ experience is as good as one with five years on the job, that is likely to evoke a spectrum of emotional responses such as jealousy or fear. Yet only 20% of CIOs said their organisation is focused on mitigating the negative impacts of AI on wellbeing, although 47% are exploring the impact of the technology on employee needs and sentiments.

All organisations implementing AI need to adopt change management and co-create employee experiences, Gartner recommends, but those in the AI-accelerated group should also hire human behaviour experts and pilot agentic AI.

Turning to financial matters, project selection is particularly important, warned Mesaglio, as “it’s really easy to waste money”. That’s partly because AI is so expensive, even for proofs-of-concept. Vendors are also raising prices as they add GenAI capabilities to their products, so “you need to understand the bill”. This is especially important for organisations on the AI-accelerated track, so they should implement real-time cost monitoring. 

Furthermore, cost estimates for GenAI can be off by 500-1,000%, said Steenstrup. So, it should be no surprise that chief financial officers are concerned by AI cost overruns, while 92% of ANZ CIOs say managing costs limits their ability to get value from AI.

Mesaglio said gaining an understanding of how costs will scale should be a part of all proofs-of-concept. Those on the AI-accelerated track also need to look for all of the benefits that may come from proposed projects, even if they are harder to estimate than productivity improvements.