Over the past few months, we’ve become accustomed to the large number of grandiose claims made about the benefits of AI. It’s been hard to escape them when so much commentary and discussion has been devoted to the subject. Finding someone who hasn’t heard of AI is probably on a par with stumbling on the Yeti.
Inevitably, those claims have drawn quite a lot of criticism because there’s a feeling the hype fuelling the vast sums of money being devoted to deliver on the promise and potential of AI is creating a very dangerous bubble. Despite the best efforts of the forced inevitability crowd to render us helpless to the inexorable advent of AI, the entries in the disadvantage column are proliferating.
I won’t go over them here again but it’s ironic that the recent report on the value of data centres to Ireland, prepared by KPMG for the Dept of Enterprise and publicised with much fanfare by Minister for Enterprise, Tourism and Employment Peter Burke should give rise to some of the growing cynicism around data centres and AI.
The latest revelation concerning the report, as detailed in The Journal, is that sections on the “opportunity cost of supporting data centres instead of housing, calls for mandated sustainability targets on water and energy use, and the impact of farmland used for data centre development” were cut from the final report. For some reason, the old adage, ‘you get what you pay for’, springs to mind.
Scaling down
Set against that, it is perhaps heartening to see an example of AI in a positive light and, it should be worth noting, on a scale that doesn’t boast of being planet shaking or world defining. According to Emergent’s Age of Custom Software Index 2026 report – based on 300 interviews with operators and a sample of more than 50,000 live applications built on its platform – a third of small business owners have built their own software to solve operational challenges that no existing off-the-shelf solution has been designed to address.
Commenting on the findings, Mukund Jha, co-founder and chief executive of Emergent, said: “We assumed we would mostly be looking at people swapping out a subscription. Instead, a third of them were solving a problem that no product had ever been built for, because their market was too small or their problem too specific for a vendor to chase. That software was never going to arrive. It only exists because the person closest to the problem finally had a way to make it.”
According to Emergent, the index demonstrates that SMEs are adopting AI tools “to produce software that addresses unique operational challenges and lowers the barrier of entry for non-technical users to design and deploy solutions”.
Up until that point, only one in seven of the 300 operators interviewed was well served by software on the market. The vast majority had to adapt with manual work, compromise on software that was not fully suitable, or pay for custom development.
AI tools have enabled the operators to build their own systems. Operators were typically people with no coding background and who had no way to build anything at all. As Emergent puts it: “Building stopped being reserved for those who could write code or afford someone who could, and moved to the person who runs the business and knows exactly what it needs.”
One of the interesting things about this story is that the operators identified where they needed an application, ascertained what they needed it to do and used AI tools to deliver it. In other words, they were able to develop custom systems suitable for their businesses which would be uneconomical for software developers to provide.
The bigger the organisation, the greater the failure
This contrasts with the experience of many larger organisations. You may remember the survey of 200 IT decision-makers in large organisations in Ireland, carried out by Censuswide on behalf of Saros Consulting in March this year.
Among the findings were that 99% had experienced AI project failure, a clear majority (57%) admitted to spending more on failed AI projects than successful ones, 53% of IT decision-makers admitted their AI made decisions they couldn’t explain to customers, 54% were unable to explain their AI’s decision-making to regulators and 53% had discovered AI making biased or discriminatory decisions in the past 12 months.
Ray Armstrong, co-founder and co-CEO of Saros Consulting, said the results showed “a pattern of businesses investing significant resources in AI, without fully understanding why, and in the hope of success that often doesn’t come to fruition”.
Maybe there’s a lesson to be learned here. While giving a speech to students at the University of Arizona in May, former Google CEO Eric Schmidt said about AI: “When someone offers you a seat on the rocket ship, you do not ask which seat. You just get on.” He was roundly booed.
What the Emergent findings show, in contrast to those of the Saros Consulting survey, is that those who choose not to view AI like a rocket ship have a better chance of arriving at their destination.
Ltd
