Why generative AI projects fail

What causes AI pilots to fail in production  

This week, an exercise in plotting truth from hype.  

I am old enough to remember when generative AI was the best thing since sliced bread – destined to solve any and all problems. But CIO.com recently reported that inconsistent results, hallucinations, and a lack of use cases tolerant to inaccuracies have sent generative AI into a downward hype cycle. 

What’s happening? Readers turned to Smart Answers for some insight. Parsing information from all our human reporting on the era of AI, our very own generative AI chatbot has thoughts. It agrees that only 16% of AI initiatives scaled enterprise-wide, and suggests that 67% of business leaders struggle to transition even half of their gen AI pilots to production. 

Reasons for this vary from pilots being designed for short-term validation, or being experiments with unclear objectives. Smart Answers cites insufficient data readiness and a lack of in-house AI expertise as causes of failure, as well as inflated expectations. 

Find out: What causes generative AI pilots to fail production? 

Productivity gains fail 

Such feelings are not restricted to the C-suite. Over in the dev team, this week InfoWorld reported that the era of cheap AI coding assistants may be over. We said that with GPU shortages and high model costs keeping prices sticky, organizations must treat vibe coding tools as a core productivity expense rather than a bargain add-on.  

But wait: vibe coding tools are a productivity win, right? RIGHT? 

Perhaps not, and readers asked Smart Answers why that is.  

As so often with AI in 2025 there may be a gap between expectation and reality. In essence, users of these tools feel more productive, but such gains may not stand up to measurement. It may also be that organizations confuse developer satisfaction with productivity, as AI tools can improve the coding experience by reducing cognitive load without necessarily leading to faster output. 

Find out: Why might perceived productivity gains from AI coding tools be wrong? 

AI reinvents business processes 

That’s the thing about hype and disillusionment – the pendulum swings both ways and truth sits in the middle.  

Back over on CIO.com one of our IT practitioner contributors told us that in their experience genAI and agentic workflows can transform business intelligence. They said that analytics dashboards are fading — GenAI and agentic AI are giving orgs real-time insights that lead straight to action, not just reports.  

Which is great. But how? By now you should recognize that Smart Answers knows.  

It says that agentic AI can drive business process reinvention by automating processes and empowering systems to act independently with minimal human intervention. Autonomous agents can reason, adapt, learn, and make decisions on complex tasks, executing end-to-end workflows. Handling routine tasks, analyzing large datasets for insights, and making preliminary decisions are all within the capabilities of AI agents.  

Smart Answers cautions that achieving true enterprise reinvention with agentic AI requires structural changes, strong leadership, and a robust data foundation to support the collaboration of autonomous AI agents in managing complex workflows. And don’t sleep in regulations. 

Find out: How can agentic AI drive business process reinvention? 

About Smart Answers 

Smart Answers is an AI-based chatbot tool designed to help you discover content, answer questions, and go deep on the topics that matter to you. Each week we send you the three most popular questions asked by our readers, and the answers Smart Answers provides. 

Developed in partnership with Miso.ai, Smart Answers draws only on editorial content from our network of trusted media brands—CIO, Computerworld, CSO, InfoWorld, and Network World—and was trained on questions that a savvy enterprise IT audience would ask. The result is a fast, efficient way for you to get more value from our content.