Serious use of AI: why 95% remains superficial and how you can change that
- Erik Hartman
- 46 minutes ago
- 4 min read
The hype surrounding artificial intelligence (AI) shows no sign of abating. The flood of exaggerated positive and negative reports about AI continues to demand attention. Take this shocking headline, for example: ‘95 per cent of GenAI pilots fail’. But is that really true?
Recent MIT research concludes that 95 per cent of AI initiatives fail and that only 5 per cent can be considered truly successful or, at least, can be developed further to the production phase within the organisation. For many people, this conclusion was reason enough to condemn AI as a failed technology.
Do not use AI superficially
But if you read the MIT research carefully, you will see that the researchers already provide the necessary nuance. Jason Snyder's Forbes article MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction on the conclusions of this MIT research also shows that the reality is more nuanced. It is not about failure, but about superficiality.
95 per cent of GenAI pilots rely on generic tools, which are good enough for demos but vulnerable in workflows. They are stuck in a mode of high acceptance and low transformation.
Seek out friction
5 per cent are designed specifically for friction. They integrate GenAI into high-value workflows, integrate deeply, and deliver tools with memory and learning loops. That's where the return on investment (ROI) lies.
The data is clear: only 5 per cent of custom GenAI tools survive the leap from pilot to production, while generic chatbots achieve an 83 per cent acceptance rate for trivial tasks but stall as soon as workflows require context and customisation.
Ensure added value
Organisations that use AI only to automate simple tasks or place chatbots on structured content are not realising the full potential of this technology. The real question is: how do you ensure that AI truly adds value?
Forbes and MIT emphasise that the difference between success and “failure” lies in the depth of the approach. The 5 per cent that do succeed do so not by avoiding friction, but by embracing it as an opportunity to learn, redesign processes and build systems that grow with the organisation.
Embrace the friction
Jason Snyder's article in Forbes emphasises that successful AI implementations actually embrace friction – resistance that learns, adapts and creates value. The 5 per cent that do succeed integrate AI into core processes, build systems that learn from feedback, and adapt workflows rather than simply adding a layer of technology. These organisations see AI not as a “quick fix,” but as a catalyst for profound change.
The MIT study confirms this: most organisations remain stuck in the “adoption without transformation” trap. They use generic tools such as ChatGPT for simple tasks, but these do not deliver structural improvement. The real gains lie in embedding AI in complex, valuable processes.
Jason describes three types of friction: human friction, organisational friction and technical friction. According to Jason, these frictions are not errors to be eliminated, but a force to be harnessed for designs using AI.
AI is more than automation
AI can do much more than just automate tasks or use chatbots for standardised content. It's about redesigning processes, learning from data, and creating adaptive systems that grow with the organisation. This requires a different approach:
Deep integration: AI must be part of core processes, not a standalone tool.
Learning and adapting: Systems must incorporate feedback, remember context, and improve themselves.
Friction as leverage: Resistance – whether technical, organisational, or human – is not an obstacle, but a signal that you are on the right track. It forces you to think about what really adds value.
Digital maturity: the foundation for success
Anyone who wants to seriously use AI for a successful digital transformation quickly realises that more is needed than just deploying the technology. At TIMAF, we know that digital maturity is essential for successful technology adoption.
Our digital maturity model helps organisations not only to implement tools, but also to develop the culture, processes and skills needed for true transformation. AI is not an end in itself, but a means to achieve strategic goals.
Want to use AI? Then you also need to tackle the processes
The trick is to know how to move from superficial AI applications to real transformation. One of those steps is to tackle the existing processes.
Our process management training teaches teams how to use AI to optimise processes, rather than just digitise them. It's about redefining workflows, measuring impact, and continuously improving processes. That's what the 5 per cent who do succeed focus on.
From superficial to impactful
How can you ensure that your organisation is among the 5 per cent of successful AI initiators?
Start with a clear goal: What do you want to achieve with AI? Is it efficiency, innovation, or customer value? Make sure AI aligns with your strategy.
Opt for deep integration: Use AI where it really makes a difference – not only in the front office (the customer layer), but also in the back office (the organisational layer), where complex processes are located.
Invest in learning and adaptation: Choose systems that incorporate feedback and improve themselves. Static tools do not deliver sustainable value.
Involve your team: AI is not an IT project, but an organisation-wide change. Ensure support and train employees to work with the new systems.
Measure what matters: Success is not measured by the number of logins, but by the number of processes that are truly improved.
Conclusion: AI deserves a serious approach
The message is clear: AI can be a game changer, but only if you take it seriously. That means looking beyond chatbots and automation and investing in systems that learn, grow and create value. At TIMAF, we help organisations take that step – from superficial to impactful.
Ask yourself: are you ready to be part of that 5 per cent?
Would you like to know more about how to use AI strategically? Contact us at info@timaf.org or take a look at our training courses.
