Many businesses rapidly embraced artificial intelligence, encouraging employees to use AI tools as much as possible, The New York Times reports.
However, many soon found that AI usage, measured in tokens, could become extremely expensive, causing companies to shift their focus from maximizing AI adoption to controlling costs.
Unlike earlier technologies such as cloud computing, AI lacks standardized pricing, transparent benchmarks and widely accepted methods for measuring its value or return on investment. Companies often struggle to determine which AI model is best suited for a task, how many tokens should be used and whether the benefits justify the expense.
To address these challenges, a new field called tokenomics has emerged to study how organizations purchase, measure and optimize AI resources. Industry groups are also working to develop common standards that improve transparency around AI pricing and performance.
Many companies now use tools to monitor token spending and compare it with business outcomes, such as software features delivered or productivity gains. Some organizations are also experimenting with treating AI as a form of virtual labor by comparing AI spending with employee costs and evaluating the combined output of human workers and AI systems.
Despite AI’s potential, uncertainty remains about its long-term economic impact. Token prices vary across providers, and while innovations such as cached tokens can reduce costs, future price changes could significantly affect AI adoption.
Another challenge is the limited number of professionals with the expertise to deploy AI effectively. Because there is little comprehensive data on AI spending and performance, businesses and economists are still trying to determine whether AI will substantially increase productivity, augment human work or become an expensive investment with uncertain returns.
The New York Times has the full story. A subscription may be required.
All-in on AI?
