Stay informed with free updates
Just sign up for artificial intelligence myFT Digest — delivered straight to your inbox.
The increasing adoption of AI in rich countries raises the risk that the technology will deepen economic inequality and widen the gap in living standards, AI startup Anthropic has warned.
An analysis of how Anthropic’s chatbot Claude is used by businesses and consumers around the world suggests that wealthier countries are more likely to adopt AI, and that there is “no evidence yet that lower-income countries will catch up.”
“Productivity gains realized in early adopters could lead to a divergence in living standards,” Peter McCrory, head of economics at Anthropic, told the FT.
The study adds to a growing body of data from tech companies showing global disparities in AI adoption rates, as well as showing that productivity benefits and business applications are concentrated in richer developed countries.
Anthropic research estimates that AI could boost annual labor productivity growth in the U.S. by 1 to 2 percentage points over the next decade, with the biggest gains occurring in complex and “knowledge-intensive” tasks.
It also suggested that about half of jobs could potentially apply AI to at least a quarter of their tasks, up from 36% of jobs a year ago.
The United States, India, Japan, United Kingdom, and South Korea use Claude the most. Brazil and the Balkans have the highest usage rates for business tasks, and Brazil is emerging as a center for legal AI. Indonesia also leads in education and coursework applications.
In low-income countries, we see a higher proportion of AI use in education and a lower number of work tasks. Claude was used more frequently for work and personal purposes in countries with higher GDP per capita, the company said.
Some content could not be loaded. Please check your internet connection or browser settings.
The study found that better-educated users derive greater productivity benefits from Claude, as they can provide more sophisticated prompts and get better responses from chatbots.
“The broad and valuable benefits of this technology may not result in sufficient adoption if left to market forces alone,” McCrory added.
Anthropic highlighted an AI literacy project it developed in collaboration with the Rwandan government. Through this project, we provided some graduates with one-year access to Claude Pro to “support their transition from educational use to broader use.”
“The ability to pay Claude’s costs varies from country to country. [educational] “The use case may be more suitable for using the free Claude than complex use cases in work areas such as software engineering,” the report states.
“Users in high-income countries may have more other resources, such as free time and continued internet access, that enable non-essential personal use cases.”
This report analyzes a random sample of 1 million conversations from Claude.ai’s free and paid conversations, consumer data, and enterprise conversations for the month of November.
This supports similar findings from Microsoft, which identified a disconnect in global AI adoption. Brad Smith, president of a big tech company, told the FT: “If we don’t address the widening AI divide, the huge economic gap between North and South is likely to persist and widen.”
