AI isn’t just one thing. It’s an umbrella term, a category of technologies and techniques¹ that encompasses a wide array of capabilities and that attempts to address a wide range of use cases. For example, while predicting customer churn, facial recognition, and robotic process automation are all legitimately “AI”, they represent three very different capabilities and AI techniques.

¹ As a handy reference, Wikipedia has an excellent and exhaustive glossary of AI terminology and concepts at https://en.wikipedia.org/wiki/Glossary_of_artificial_intelligence

Three recent advancements have pushed AI to the forefront:

• Supporting technologies and computing power have progressed to make AI techniques achievable and widely applicable for business and consumer needs.
• As AI techniques have evolved and become more powerful and accessible, new and novel use cases have been identified. Uber doesn’t exist without a brilliant idea and AI to make it a reality.
• Most importantly, today’s digital environment generates vast amounts of readily and quickly available data that, in turn, power and enable AI solutions.

AI projects share many familiar failure points with non-AI projects – scope creep, insufficient or missing requirements, unrealistic timelines, insufficient resources, lack of user input, inadequate testing, insufficient communication and training, lack of user adoption, insufficient post-implementation support, and many more. While these are all important concerns, AI projects are most susceptible to:

• Unrealistic expectations – When addressing some business challenges, AI can be extremely powerful, but its not magic. Businesses should carefully evaluate potential value and costs before making an investment.
• Unanticipated costs – AI investment requirements extend far beyond technology and implementation. You’ll also need to factor in the cost of data infrastructure; data acquisition(internal and external); ongoing testing, validation, and retraining; careful curation of data/model/result sets; compliance requirements for explainability and ethical behavior; and change readiness, stakeholder management, communications, training, and reinforcement.
• Solution viability – Of all failure points, solution viability is perhaps the most ambiguous and critical. For some business problems, AI just isn’t a viable approach.
• User acceptance – The premise and promise of AI means that some people’s jobs are going to change. When users can’t accept and support the changes that AI brings, projects fail.