Just because you automate a process doesn’t mean you’ve secured it. If you're considering RPA, make sure you understand the security implications
8 reasons AI projects fail
Why do artificial intelligence projects fail? Let’s examine some recurring issues – and expert advice on how to avoid them and increase your AI project’s chances for success
It’s a sobering stat: Seven out of 10 executives whose companies had made investments in artificial intelligence (AI) said they had seen minimal or no impact from them, according to the 2019 MIT SMR-BCG Artificial Intelligence Global Executive Study and Research Report.
At the heart of the matter may be a general lack of understanding about AI capabilities and requirements. “At this point in time, many enterprises have inflated expectations from AI solutions,” says Anil Vijayan, vice president at Everest Group. “This can often create a mismatch between what is expected and what is achievable.”
But “get smarter about AI” is not the most nuanced takeaway. In fact, there tend to be some more specific recurring reasons why AI projects fail – and steps IT leaders can take to increase their chances for success. Here are eight of the most common mistakes and miscalculations that can portend AI project failure.
[ Check out our quick-scan primer on 10 key artificial intelligence terms for IT and business leaders: Cheat sheet: AI glossary. ]
8 causes of AI project failure
1. Shiny things disease
“Most digital journeys begin with a technology-first orientation, going deep into a solution’s capabilities [such as] confirming which machine learning libraries it uses,” says JP Baritugo, director at business transformation and outsourcing consultancy Pace Harmon. “Instead, firms should first concisely articulate the key business imperatives it wants to address. Once defined, these objectives then drive and inform what digital and transformational interventions to pursue – including AI.”
Many leaders may be unclear about where AI will be best leveraged in their organization. Working closely with the business to identify where AI might solve an existing problem or focusing on areas where others have found AI to be valuable – like marketing, financial planning, or risk analysis – can be good places to start.
2. Insufficient training data
“AI solutions do require meaningful, labeled training data sets to achieve desired outcomes,” says Vijayan. “Often, lack of availability of data for training is a key reason for failure.” Depending on the type of AI being applied, that may mean anywhere from thousands to millions of data examples to train the model.
3. Poor data governance
Garbage in, garbage out still applies. “Most enterprises underestimate the importance of quality data in enabling AI implementation success,” Baritugo says. “Unfortunately, some companies have poor data governance and poor data hygiene practices that result in data that are suspect, duplicated, or called something else somewhere else. Moreover, these firms have multiple, disparate systems housing bits and pieces of the required information.”
Risks can also arise from skewed data samples. “This can lead to problems such as overfitting, for instance, thereby leading to incorrect outputs when run in production,” says Vijayan. “AI systems will only learn what they are fed. So there’s always a risk of human biases being learned and propagated through machines.”
[ How can you guard against AI bias? Read also AI bias: 9 questions for IT leaders to ask. ]
Establishing master data management and governance and developing a central data repository (a data engagement platform or data lake) is mandatory. “To create transformative AI solutions, we need a holistic, synergistic, and simultaneously integrated flow of information,” says Seth Earley, author of “The AI-Powered Enterprise: Harness the Power of Ontologies to Make Your Business Smarter, Faster, and More Profitable.” Earley says that a consistent representation of data and data relationships that can inform and power AI is “the master knowledge scaffolding” for AI-driven transformation.
4. Underestimating the related culture change
According to the Gartner report Predicts 2020: Artificial Intelligence – The Road to Production: “The cultural impact of introducing AI to both customers and employees has largely been underestimated. A new dynamic between humans and machines is being teased out with the realization that AI takes tasks from employees, not jobs. Employees are adapting to new roles as exception handlers and trainers, often working alongside AI systems in a symbiosis we call augmented intelligence.”
Consider creating programs to increase employee understanding and skills and increase acceptance and engagement. “AI projects could fundamentally alter how work is performed and how decisions are made,” Pace Harmon's Baritugo notes. “Without well-thought-out change management efforts, the business users may not accept the AI results (or view them with suspicion) and dampen the overall AI adoption.”
Let’s examine 4 more common factors that can cause AI projects to fail: