The collective illusion that one only has to add ‘AI‘ to a strategy to make it successful is collapsing under the weight of numbers. According to a new study published by MIT and relaunched by Fortune, 95 per cent of AI-based pilot projects fail before they reach the real adoption stage. To read the data, one would be forgiven for thinking of a technological error. But the problem is not the code: it is the management.
The report debunks the idea that algorithms, GPUs or lack of datasets are to blame. AI projects fail because the wrong targets are chosen, the internal skills to interpret them are lacking and, above all, the company management is not prepared to manage the structural transformation that artificial intelligence inevitably entails. There is much talk of generative AI, but when it comes to implementing concrete solutions in production, financial or operational processes, many companies get stuck in an experimental stage without vision or governance.
The systemic error, highlighted by MIT, is not the technology but the approach. The C-suite often seems attracted by the allure of demos, promises of instant automation or slides of enthusiastic consultants, but ignores the fact that AI is not a tool to be installed: it is a paradigm shift that requires reorganisation, data literacy and a deep understanding of the inner workings of models. In essence, there is a lack of true AI literacy at the decision-making level.
In this scenario, investors are starting to become more cautious. The market, which had enthusiastically embraced any startup with ‘AI’ in its name, is beginning to demand tangible results, concrete use cases and metrics that go beyond vanity. It is a trend reminiscent of the parable of cryptocurrencies: even then, after the initial phase of euphoria, only the truly structured projects have endured. Interestingly, solid realities such as Bitcoin or Ethereum have withstood the shock of the markets precisely because of their long-term vision, despite volatility and cyclical crises of confidence.
Those working with artificial intelligence today face a crossroads: ride the hype narrative or build solid foundations. Success stories exist, but they are few and far between, often within companies that have invested in staff training, process revision and the construction of truly scalable digital architectures. That is where the real game is played. AI will not replace managers, but it will make those who do not understand it irrelevant.
The most disturbing fact of the MIT report is not the failure rate. It is the fact that many companies keep repeating the same mistakes, convinced that it is enough to buy an LLM to enter the future. But the future cannot be bought. It is built, brick by brick. And yes, sometimes even in reality.
Here is the source: https://fortune.com/2025/08/21/an-mit-report-that-95-of-ai-pilots-fail-spooked-investors-but-the-reason-why-those-pilots-failed-is-what-should-make-the-c-suite-anxious/
