An artificial intelligence (AI) search company called Lucidworks recently published a report on what they characterized as a slowdown in AI spending, and it brought to mind a roundtable discussion at our recent Future of Field Service Live event in Cologne, Germany.
During my final interview with Jelle Coppens, Product Domain Expert for Service and Repair at Electrolux, the discussion turned to AI. Electrolux is using IFS's AI-powered planning and scheduling optimization engine (PSO) and plans to expand their use of the technology. The conversation was a great real-world use case of AI in action, but what struck me was that during the roundtable sessions after the session, some participants said they are growing tired of hearing about AI.
I get it. AI is an inescapable topic, no matter what industry you are in. Much of the roundtable venting had to do with the volume of AI coverage and the lack of specifics on exactly where and how to use the technology. (“We are tired of hearing about this technology” is not part of the Gartner hype cycle, but maybe it should be!)
That's why the Lucidworks report caught my eye, because their data indicates actual deployment experiences are throwing some cold water on AI-mania, and that will probably help make the projects that do move forward a lot more successful. I think what has happened is that the bandwagon took off without many defining a clear business case or selecting proven, functional tools. Now we are stepping back to assess how best to make use of AI, and I can see how all of the conversation can cause some to grow weary.
However, I do believe AI is an incredible opportunity – in field service and beyond. One we need to take caution to harness appropriately and balance with humanity, but the potential to layer more intelligence into existing digital ecosystems is massive.
AI in Field Service
In field service, AI is (at least near term) best suited for what the tech industry now refers to as co-pilot scenarios, where the algorithm exists to enhance or augment workflows, rather than supplanting the real humans doing that work. In applications where there are simply too many variables or too much data for a person to possibly evaluate accurately, AI can help narrow choices and point people in the right direction. It can also help to automate work that is time consuming but low value, and to heighten predictive capabilities.
AI projects that are not well planned or properly implemented can quickly prove to be costly and useless. Large language models (LLMs) trained on unfiltered data can hallucinate, and models that ingest AI-created data can suffer from what is known as model collapse (I talked about this back in April).
The Lucidworks survey indicates that AI adoption is beginning to slow because of some of these concerns, with just 63% of organizations planning to increase AI spending this year, compared to 93% in 2023. According to the study, just around one quarter of planned projects are fully implemented, and 42% have not produced significant benefits. In many cases, projects haven’t made it out of the pilot stage.
The number of companies worried about project cost has gone up 14 times compared to 2023, and concerns about response accuracy have increased by a factor of five, The more complex the application, the more these concerns increase (along with costs), while success is harder to achieve.
The WBR Insights "AI in Field Service Report" also mirrors some of the Lucidworks findings. In that survey, 92% of respondents said they struggled with legacy integration in their AI implementations, and 74% were challenged by a lack of data quantity or quality. Costs were a problem for 62% of organizations.
Most of the companies in the WBR were already using AI for predictive maintenance (88%), parts wastage prevention (82%), case predictions (58%), and call deflection (57%).
When it is deployed successfully, AI can produce notable benefits in time savings, increased first-time fix rates, reduced parts wastage, and faster resolution times. For instance, another report from MarketsandMarkets claims the integration of mobility and AI in field service can result in 30% to 40% productivity gains.
So, are we facing AI fatigue? In some cases, yes – but not because the technology is overhyped. Rather because many companies leapt into AI projects haphazardly when the buzz began, only to learn the business benefits of AI demand a much more fastidious approach. To me that’s what this data represents; a collective step back to take a more tempered, strategic approach to AI, which will ultimately pay off.