To front page
Contact us
Why AI projects Fail

Why 95% of AI projects fail, and how industrial companies can be among the winners

Only 5% of AI pilots deliver measurable business value. Discover how industrial companies can succeed not through technology alone, but by transforming how people and processes work alongside AI.

There is plenty of hype around generative AI, but one number stops you in your tracks: according to an MIT report (Opens in new window) published earlier this summer, 95% of GenAI pilots fail to deliver measurable business value. Rather than technology, the reason usually lies in managing the transformation: companies fail to integrate AI into their everyday life and workflows, where its value would be greatest.

For industrial companies, the stakes of failed projects are even higher: production efficiency, productivity, safety, quality, and supply chains are all mission-critical. Investing in technology is necessary, but it is equally important to ensure that solutions are actually used and that ways of working evolve with the technology. After all, technology can generate real business impact only when it is actually being used.

Why AI projects fail – and what to do differently

  • AI doesn’t have a role in the user’s life : If a solution doesn’t fit into existing workflows or help the user with their daily tasks, it stays disconnected and unused. That is why users need to be involved from the very beginning to create shared ownership for future work: technology + processes + people.
  • AI user experience is ignored: Trust disappears quickly if AI feels like a black box, with its inner logic carefully hidden. To build trust, tools and systems must be designed so that users can easily understand, accept, and override AI decisions when necessary.
  • Regulation – Minimum compliance or competitive advantage: The EU AI Act requires human oversight for high-risk AI systems. While this is a regulatory requirement, smooth human-AI collaboration also reduces business risk, improves usability, and simplifies adoption – even in systems that aren’t classified as high risk.
  • AI is measured with wrong goals and metrics: Simply “experimenting with AI” is not enough. Success must be measured on two levels:
    • AI Solution KPIs: model accuracy, explainability, and user trust.
      Tracking these metrics throughout iterative development shows whether the solution works technically and predicts how well the users will adopt it.
    • Business KPIs: production efficiency, quality, accuracy, or employee experience. These show whether the AI project achieves a true cost–benefit ratio. If the cost–benefit ratio is not identified or monitored, there is a risk that the AI solution will remain an isolated technical experiment without real business impact.

Practical steps to thriving AI projects

  1. Define the use case and clarify value: choose a problem with a significant, measurable business impact aligned with your strategic priorities.
  2. Pilot with users: involve operators and experts in the design to ensure the solution fits the daily work of its users in a meaningful way.
  3. Keep the human in the loop: apply human-in-the-loop (keeping a human in control of key decisions) principles even when not strictly required – it builds trust and drives adoption. 
  4. Ensure skills and data readiness: make sure your team has expertise across technology, process, and human factors – and that your data quality supports continuous improvement.
  5. Think long term: developing AI solutions is a marathon, not a sprint. Refining language models takes time – treat it as a natural part of the journey rather than expecting instant wins.

If these things are in place, your next AI project can be among those that succeed.

Ready to turn AI potential into real results? Connect with our AI expert today!

About the author

Hanna Remula

Head of Design, Cloud and Applications

Hanna Remula is a business developer and design leader with a passion for driving meaningful transformation in industrial companies. With deep expertise in strategic design and change adoption, Remula helps organizations go beyond technology – ensuring that digital, data, and AI initiatives deliver genuine value. Remula bridges the gap between business, technology, and people across the OT–IT landscape to enterprise-wide digital, data and AI programs to deliver real impact. She is a trusted partner to industry leaders, SMEs, and global corporations shaping the future of industrial operations.

Ask our expert a question

Hanna Remula

Head of Design, Cloud and Applications