A report claims that, apart from the Mag 7 (the seven tech giants), AI has yet to deliver any financial returns to other U.S. companies.
Its evidence: companies outside the Mag 7 have seen no significant growth in profit margins or revenue per employee.
But there’s a key problem with this conclusion: a lack of growth in aggregate data doesn’t mean AI hasn’t created value.
A more reasonable way to measure AI’s contribution is to compare companies that use AI to different degrees: how do companies that use more AI differ from those that use less in productivity, costs, and profits?
Looking only at the overall average across all companies makes it difficult to distinguish AI’s actual contribution. Factors such as interest rates, energy costs, wages, and changes in demand can all affect the final results.
The way the companies are grouped is also debatable. The Mag 7 aren’t just AI users; they’re also major suppliers of cloud services, chips, models, and AI infrastructure. Many other companies, by contrast, are primarily buyers of AI. The two groups already have different business models and revenue structures, so directly comparing their profit margins makes it difficult to isolate AI’s effects.
The original article also mentions the economic pressures at the time: rising interest rates, oil prices reaching $116 a barrel, high inflation, and consumer confidence at a historic low.
If companies can still maintain their profit margins in such an environment, it’s worth investigating whether AI helped them offset some of the cost pressures. But that alone doesn’t prove AI has already delivered returns; more detailed data is needed.
So the issue isn’t that we should assume AI must be effective. It’s that we shouldn’t use an aggregate metric that can’t identify cause and effect to declare that AI has no financial value.
Aggregate data can tell us what happened, but not necessarily why. To measure AI’s contribution, we need to separate it from other influencing factors.
Its evidence: companies outside the Mag 7 have seen no significant growth in profit margins or revenue per employee.
But there’s a key problem with this conclusion: a lack of growth in aggregate data doesn’t mean AI hasn’t created value.
A more reasonable way to measure AI’s contribution is to compare companies that use AI to different degrees: how do companies that use more AI differ from those that use less in productivity, costs, and profits?
Looking only at the overall average across all companies makes it difficult to distinguish AI’s actual contribution. Factors such as interest rates, energy costs, wages, and changes in demand can all affect the final results.
The way the companies are grouped is also debatable. The Mag 7 aren’t just AI users; they’re also major suppliers of cloud services, chips, models, and AI infrastructure. Many other companies, by contrast, are primarily buyers of AI. The two groups already have different business models and revenue structures, so directly comparing their profit margins makes it difficult to isolate AI’s effects.
The original article also mentions the economic pressures at the time: rising interest rates, oil prices reaching $116 a barrel, high inflation, and consumer confidence at a historic low.
If companies can still maintain their profit margins in such an environment, it’s worth investigating whether AI helped them offset some of the cost pressures. But that alone doesn’t prove AI has already delivered returns; more detailed data is needed.
So the issue isn’t that we should assume AI must be effective. It’s that we shouldn’t use an aggregate metric that can’t identify cause and effect to declare that AI has no financial value.
Aggregate data can tell us what happened, but not necessarily why. To measure AI’s contribution, we need to separate it from other influencing factors.