95% of AI pilot projects fail – SA businesses must act wisely

A recent MIT report featured in Fortune magazine revealed that 95% of generative-AI pilot projects are failing, resulting in investors pulling back and misrepresenting this as a failure of the technology itself.

However, the real failure lies in the way organisations are adopting and managing it. Artificial intelligence does not break business – a poor strategy does. The problem is not how AI has failed to deliver, but how too many companies have rushed into it without a plan.

“Too many organisations start with a solution and then look for a problem. This is a wholly redundant approach,” says Daniel Novitzkas, chairperson of Specno. “Instead, it should be the reverse: identifying a hurdle and assessing whether AI can realistically mitigate it. This requires clarity on workflows, data quality and change-management processes – the same fundamentals that determine success in any complex software project.

“The MIT study shows that generic, off-the-shelf AI tools may be adequate for individuals, but regularly fail in complex business environments. By contrast, bespoke AI tools designed by specialists enjoy a 67% success rate, far higher than the 33% success rate of internally built tools. The lesson is clear: Technology must adapt to the business, not the other way around,” he adds.

For South African businesses, the imperative is not whether to adopt AI, but how. The country faces slow economic growth, collapsing infrastructure and intensifying competition from global giants such as Amazon and Walmart. In such an environment, efficiency is not optional but rather essential. AI offers a practical way to do more with less: automate repetitive tasks, improve forecasting accuracy and enrich customer insights.

Novitzkas clarifies that “caution is naturally healthy, but fear should not invalidate action. AI’s value rarely arrives with fanfare; it appears quietly in shorter service queues, smarter inventory management and fewer compliance errors – the kinds of improvements that build resilience over time. When leaders see AI as a tool to enhance core processes, the results are tangible: saved time, reduced costs and accelerated development.

“Practical examples from technology partners such as Specno demonstrate what deliberate integration can achieve. Using AI in product development and validation, Specno and its clients have cut validation cycles significantly, while reducing project costs by up to 30%. By embedding AI into traditional workflows and automating design, testing and analysis, companies are delivering faster, more efficiently and with greater confidence,” he continues.

“It has never been easier for companies to get the right help. The strategic deployment of AI in the early stages of developing a minimum viable product enables a shorter feedback loop, allowing businesses to make critical decisions more quickly. This faster turnaround means more time dedicated to input and iteration, resulting in stronger, more reliable systems that minimise disruption to operations.”

Still, many businesses treat AI as a shiny add-on rather than embedding it into their core operations. This adoption gap remains one of the biggest barriers to success. Technology cannot compensate for the absence of strategy, governance or skilled people. Sustainable adoption requires all three: clear direction, capable teams and thoughtful process design.

“AI is not a magic bullet. It is objectively a tool. When integrated with clarity and purpose, it has the power to transform efficiency, sustainability and competitiveness. The future is already rewarding those who choose to make it work intelligently,” Novitzkas concludes.