If you’re tuned in to what’s happening in the AI world, you’ve probably come across the recent MIT report showing that 95% of all AI initiatives in the enterprise world fail. Sweden and the Nordics are also far behind when it comes to GenAI investments. Almost everything that happens stops at the PoC or pilot stage. The result? Instead of getting real bang for the buck, we’re left with a growing graveyard of pilot solutions that never take off – yet still cost money to maintain.
So why is this happening?
Technology evolves faster than culture.
Organizations invest 90% in algorithms and tech but only 10% in people and processes. That’s often where AI initiatives die, because AI, like so many other things, isn’t an IT project. It’s a change initiative, and that means the organization must have both the ability and the willingness to change.
Leadership and IT are pulling in different directions.
Leadership wants to show initiative and shouts “Go!” while IT and security are holding the brakes, asking whether they even can or should let AI in. It’s hard to build and scale something that isn’t sanctioned and supported across the entire organization.
Business value is forgotten.
Surprisingly many projects start without clear use cases or business cases. That makes it difficult to get leadership buy-in, or to even know whether the solution creates any real value.
Yet what struck me most was the tone of the discussions. There was a surprising amount of talk about failures, poor ROI, and missing value. And while I do sometimes wish for a more sober perspective on the “AI revolution,” I don’t entirely agree that the lack of major, production-level success stories is particularly strange at this stage.
Generative AI – not to mention the potential emergence of AGI – represents a massive paradigm shift, introducing entirely new ways and tools to get things done. Every major technological leap in history – electricity, the internet, mobility – has required an exploratory phase.
That phase where we don’t know much yet. Where everything is wow, hype, experiments, and often empty slogans like “We’re shaping the future,” “We’re in the driver’s seat of the AI revolution,” or “AI is in our DNA.” But without that phase, we never reach the next one. This is where we learn – where we allow ourselves to fail fast and big so we can later build things that are bigger and better.
The key is not to confuse exploration with delivery – or bragging with substance.
If we want to move into the next phase, one where we avoid yet another generation of AI zombies and instead build sustainable AI solutions that truly deliver value and return, we need to start talking more about change management, processes, and business impact – and a little less about the technology itself.
