Using AI might lead to some improvements, but it hasn’t led to a profit revolution.
AI might be the big buzzword that’s driving hundreds of billions in global datacenter and chip fabrication development, but no one has quite figured out how to make it profitable yet. Questions are now arising about whether they’ll ever manage it, too, as a new study out of MIT sends investors into a tizzy. It found that 95% of businesses that implemented AI in their day-to-day operations did not end up with a big uplift in profits.
Over the past few months, more and more calls have started to come in that the AI industry is looking like bubbles of the tech industry’s past. The Dotcom boom (and bust) is perhaps the best example, but similar hype trains like blockchain technology, Web 3.0, and NFTs have all been cited as potential comparisons. Although major tech leaders like Sam Altman, Elon Musk, and Mark Zuckerberg might claim we’re on the cusp of AI changing the world, it may not do much for real businesses.
The MIT study, The GenAI Divide (you can request access to it here), examined 300 public AI implementations and interviewed hundreds of employees and business leaders. It found that just 5% of the companies investigated saw a large revenue growth. In comparison, the rest demonstrated little to no benefit of using AI when it came to increasing business income and profit.
Although many reporters and analysts have called the AI boom a bubble (even Altman has suggested it may be overinflated), this report seemed to strike a nerve where others haven’t. Tech stock investors responded immediately, with notable drops in global prices for big names like Nvidia, ARM, Oracle, AMD, and Palantir.
The glass-half-full response to this study is that some companies have seen tremendous revenue growth when adopting AI. Report lead author, Aditya Challapally, told Fortune that where many companies try to build their own AI, companies that purchased or subscribed to an existing model were far more successful.
“Almost everywhere we went, enterprises were trying to build their own tool,” he said, but by the data, buying a pre-built solution was often far more effective.
Other commonalities with successful companies included letting staff and low-level managers have more choice in the AI they used and in how it was implemented. Companies and start-ups led by younger people in their late teens and early 20s tended to perform better, too, with some going from “$0 to $20 million in a year,” though that was far from the norm in studied businesses.
Tech stocks have largely recovered since the news, and the AI hype train continues to chug along, but the foundations of what makes the AI revolution actually worthwhile have been shaken. If AI continues to remain impressive but with limited avenues towards profitability, the massive boom of AI investment may turn towards a bust before too long.
