Will generative AI pay off for businesses in 2024?

Underwhelming returns on AI investments led many businesses to believe the tech is overhyped

Many businesses that integrated generative artificial intelligence tools into their operations last year have not seen the returns they expected, leading some to believe the new technology is over-hyped. One expert says this is the year those businesses will be demanding results.

"2024 is the year businesses are saying ‘show me the money’ from AI" says Arijit Sengupta, CEO and founder of AI app developer Aible and creator of Harvard Business School’s course "AI in Market Facing Functions." 

woman typing holding AI symbols

The average ROI on generative AI projects last year was not great, according to a study conducted by IBM. (iStock / iStock)

"CFOs want AI results now, yet ROI continues to be out of grasp for many organizations despite the success of early adopters with AI experience," Sengupta said.

The former Harvard Business School executive fellow pointed to a study by IBM showing the average return on investment for Gen AI projects last year was just 5.9%, which the analysts noted was "well below the typical 10% cost of capital."

WHITE HOUSE: DEVELOPERS OF ‘POWERFUL AI SYSTEMS’ NOW HAVE TO REPORT SAFETY TEST RESULTS TO GOVERNMENT

Sengupta said the data highlights a crucial gap between the theoretical promise of AI and its practical implementation in the business world.

illustration of businessman holding AI in hand

A businessman touching the service AI-powered chatbot, analyze customer data, business and technology. (Shutthiphong Chandaeng/iStock / iStock)

"We are already in an AI bubble in terms of the disconnect between user expectations being driven by aggressive marketing, and what the technology can deliver out of the box," he told FOX Business.

SHOULD BUSINESSES DISCLOSE THEIR USE OF AI? MOST AMERICANS SAY YES

According to Sengupta, ROI continues to be out of grasp for many organizations because lead times are too long. He says that if implementation takes longer than a month, the underlying technology becomes outdated before the project is done and argues "projects should take hours, days, or weeks at most."

Instead, many companies are waiting too long to have what they deem to be "good enough" data before implementation. "Perfect data is a pipe dream," Sengupta says. "Prioritize getting the tech in the hands of end users quickly, clean it based on their feedback, and the AI will improve as you go."

GET FOX BUSINESS ON THE GO BY CLICKING HERE

"It takes a little bit of configuration and context transfer to make AI shine," he said. "But the marketing doesn't talk about that reality, thus leading to customer dissatisfaction."