Study Suggests Ways to Overcome High Failure Rate in AI Projects

While investors and companies around the world are excited about the use of artificial intelligence (AI), the outcomes of AI projects are often poor.

A Rand Corporation study published this month noted that: “By some estimates, more than 80% of AI projects fail — twice the rate of failure for information technology projects that do not involve AI.”

The three authors of the study, entitled, ‘The Root Causes of Failure for Artificial Intelligence Projects and How They Can Succeed’, said they interviewed dozens of data scientists and engineers with at least five years of experience in AI and machine learning models to find why so many projects fail – and how they can help make AI projects more likely to succeed.

ALSO SEE: 40% of Global Energy Generated by Renewables in 2023

They listed five root causes why AI projects failed, saying industry stakeholders often misunderstand — or miscommunicate — what problem needs to be solved with AI and that many AI projects fail because the organization lacks the necessary data to adequately train an effective AI model.

“In some cases, AI projects fail because the organization focuses more on using the latest and greatest technology than on solving real problems for their intended users,” they said, adding that organizations might not have adequate infrastructure to manage their data and deploy completed AI models, which increases the likelihood of failure.

And finally, they said, some “AI projects fail because the technology is applied to problems that are too difficult for AI to solve”.

So what can people do?

The study, which aims to “avoid the anti-patterns of AI”, gave a list of recommendations.

Industry leaders should ensure technical staff understand the purpose of a project and its context, they said, as misunderstanding or miscommunication about the intent and purpose of a project are the most common reasons why an AI project fails.

Industry leaders should select enduring problems, because AI projects require time and patience to complete.

“Before they begin any AI project, leaders should be prepared to commit each product team to solving a specific problem for at least a year.”

Industry leaders should focus on the problem themselves, because “successful projects are laser-focused on the problem to be solved, not the technology used to solve it.”

They should also invest in infrastructure to support data governance and model deployment and cut the time required to complete AI projects to increase the volume of high-quality data available to train effective AI models.”

Business people also need to understand AI’s limitations and be able to assess the feasibility of a project.

And academics should develop partnerships with government to overcome data collection barriers and give researchers access to data needed for academic research.