Artificial Intelligence: How the Labor Market Will Change Due to the Rapid Adoption of AI.

According to Vox: 1956. You are a researcher working for the International Business Machines Company, a well-known manufacturer of tabulating machines that has recently entered a new segment – electronic computers. Your task is to understand how your customers are using the large IBM EMs.

The Use of Computers for Military Purposes

The answer turns out to be very simple: Computers are primarily used for military purposes. In 1955, the main source of revenue for IBM's computer division was the SAGE project, an initiative of the Defense Department that required IBM to develop a computer alert system for early warnings of potential nuclear attacks by Soviet bombers. This generated $47 million in revenue, while other military projects added another $35 million. At the same time, programmable computers sold to companies generated only $12 million.

You send a memo to your supervisor explaining that the impact of computers on society is likely to provide the USA with an advantage over the USSR during the Cold War, while their impact on the private sector seems insignificant. Leaning back in your chair with a cigarette in hand, you ponder the prospects of the military-industrial complex.

Significant Changes in Revenue

However, you would undoubtedly be wrong – not only about the distant future but also the near one. Two years after 1956, sales of programmable computers to private companies matched those of the SAGE project. The following year, the private sector began generating as much revenue as the military. By 1963, less than a decade after your observations in 1955, military revenues appeared as a negligible addition to the rising revenues of IBM from private computers, which already constituted the majority of total sales for the company in the USA.

Who would have expected that!

What Lessons Can We Draw from Today's Use of AI?

This week, teams of economists from OpenAI and Anthropic presented large, carefully crafted reports on how people are using their AI models. The first thought was: 'It would be interesting to know what IBM's report on the use of the first computers would have looked like.' (Note: Vox Media is one of several publishers that have signed partnership agreements with OpenAI. Our journalism remains editorially independent.)

Note: The level of care that AI teams are investing in their work is significantly higher than what the fictional IBM analyst demonstrated. Revenue is not the best indicator of actual interest and usage; in 1955, everyone knew that computers were going to get better quickly, and their usage would change. AI companies have access to an impressive amount of real-time data about how their products are being used.

The Growth of AI Popularity

• Usage is growing rapidly: The number of registered ChatGPT users has increased from 1 million in December 2022 to 100 million in November 2023, and currently, there are over 750 million requests per week. If the number of requests continues to grow this quickly, by the end of next year there will be more requests to ChatGPT than search queries on Google.
• Both companies have found that AI is used more frequently in wealthy countries, but OpenAI also notes that middle-income countries like Brazil use ChatGPT almost as actively as wealthy countries like the USA.
• Key use cases for ChatGPT include 'practical tips', such as learning and tutoring (28.3% of requests), editing and translation (28.1%), and information inquiries (21.3%). Users of Claude.ai most commonly use it for solving mathematical problems (36.9% of usage).

What Can't We Learn?

But I am relentless. I am not only seeking to know the first descriptive facts about the use of these models, although precisely these questions could be addressed in the mentioned reports. I am most interested in the unconventional use of AI and its economic consequences, particularly:

• Will humans and AI complement each other, or will they replace each other in five, ten, or twenty years?
• Will salaries rise because the economy still depends on human labor? Or will they fall to zero, as such constraints do not exist?
• Can AI produce 'genius in data centers' that conduct their own scientific research? Will this lead to a jump in scientific knowledge about the world?

Many people are already asking these questions, and numerous theoretical research papers have already been conducted on this topic. I know a useful collection of lectures and articles by economist Philip Trammell.

Conclusion

It is no less important to draw attention to what OpenAI and Anthropic confirm: The spread of AI is happening faster than previous technologies. Like other global technologies, this will avoid delays that we faced in the past. We probably will not have the same time for adjustment.

Insights into the rapid development of artificial intelligence technologies and their impact on various aspects of the economy are of utmost importance. Observing how AI transforms approaches in education, business, and science can offer new perspectives for analyzing societal changes and challenges we will face in the near future.