Why AI Projects Fail: Yonathan Cohen Highlights the Pitfalls of Hasty Implementation

In a recent LinkedIn post, Yonathan Cohen discusses a critical reason behind the widespread failure of Artificial Intelligence projects: the tendency to implement before truly understanding the underlying business problem. Cohen, a proponent of a more strategic approach, argues that many organizations leap into AI adoption without a clear objective, leading to wasted resources and disillusionment with the technology itself.

Cohen highlights a common, yet detrimental, pattern: a company hears about AI, feels compelled to implement it, but lacks a defined problem to solve. This often results in projects that ultimately fail, leading to the incorrect conclusion that “AI doesn’t work.” He emphasizes that the technology itself is rarely the issue, but rather the approach taken by the implementing organization.

“80% of AI projects fail. Because of this one thing: We implement before we understand.”

The Common Mistakes in AI Implementation

Yonathan Cohen outlines several key mistakes that derail AI initiatives. One of the most significant, according to Cohen, is the pursuit of a solution without a clearly identified problem. He contrasts the vague notion of “We need AI” with a more effective starting point: identifying a specific pain point, such as “We waste 20 hours on X.” This reframing, as Cohen suggests, shifts the focus from the technology to the business need.

Another critical error identified by Cohen is the lack of defined success metrics. Without a clear understanding of what constitutes a successful outcome, it becomes impossible to measure progress or determine if the AI implementation has actually achieved its goals. As Cohen puts it:

“No success metrics. Can’t even define what ‘working’ looks like.”

Furthermore, Cohen points to the common pitfall of attempting to do too much too soon. The ambition to “Revolutionize everything” often leads to achieving nothing concrete. This broad, unmanageable scope can overwhelm resources and obscure the path to tangible results.

The Path to Successful AI Implementation

In contrast to the common failures, Yonathan Cohen details a simpler, more effective approach taken by successful AI projects. He argues that these initiatives typically begin with a focused strategy:

A Focused, Incremental Approach

According to Yonathan Cohen, the key to successful AI implementation lies in starting small and being highly specific. The winning formula, as he describes it, is:

“One small problem, one clear metric, one workflow. That’s it.”

This methodical, step-by-step process allows organizations to build momentum, demonstrate value, and learn from each iteration. By concentrating on a single, well-defined problem with measurable outcomes and a streamlined workflow, businesses can increase their chances of AI success. Cohen’s analysis suggests that a disciplined, problem-first approach, rather than a technology-first rush, is the true differentiator between AI project success and failure.

Yonathan Cohen’s insights serve as a valuable reminder for businesses looking to leverage AI: understand your problem before you seek a solution, define what success looks like, and start with manageable steps. As Cohen concludes, it’s the strategic approach, not the AI itself, that determines the outcome.