More than 80% of AI projects fail, study reveals

A artificial intelligence (AI) is one of the most popular topics among investors looking for the “next big thing.” However, a survey by RAND Corporation indicates that over 80% of AI projects fail, double the failure rate of non-AI tech startups. Think Tank Global surveyed 65 data scientists and engineers with experience in the AI ​​industry and identified several reasons for this high failure rate.

One of the main reasons why AI projects fail is the disconnect between leaders’ goals and practical reality. Expectations are often based on Hollywood-influenced misconceptions about AI rather than on technical reality. This leads to a lack of resources and time needed to achieve the set goals.

Read also
78% of Brazilian startups already use Artificial Intelligence in their processes
NVIDIA is a bubble and AI is overhyped, says investment fund Elliott

AI engineers themselves also contribute to the problem. They can get carried away by the latest advances in AI, applying them to projects without considering their true value. This “shiny object syndrome” causes professionals to want to use new technologies simply because they are new, without considering whether they effectively solve existing problems or simply add complexity.

Other factors include a lack of suitable data sets, poor infrastructure, and a mismatch between AI and the problem being solved. The research also points out that these challenges are not unique to the private sector; academia also faces challenges, focusing more on publishing research than on finding practical applications for its results.

These findings explain the many bankruptcies and consolidations seen in the AI ​​industry. Baidu CEO Robin Li Yanhong noted that China has a glut of large language models with few practical applications, wasting significant resources. Despite China filing six times more generative AI patents than the US in the past decade, only the Chinese Academy of Sciences is among the top 20 most cited entities from 2010 to 2023.

Companies are rushing to lead the way in AI, but it would be wise to learn from the mistakes of other AI projects. If projects continue to fail to deliver on their promises, the entire industry could collapse like a trillion-dollar bubble.