[ad_1]
The future of AI, depending on whom you ask, is everything and anything. It is the key to never-ending economic growth. It is the cause of catastrophic unemployment. It is the driver of lifesaving medical advances. It is the source of grave risks to national security.
The truth is that no one knows the full implications of AI. But what researchers like us are starting to see in the data is that it isn’t the speed of AI development that will shape the near-term future. It is the speed of its adoption. And AI-use across various sectors show that adoption is uneven and slower than the Silicon Valley and Wall Street hype suggests.
In August, a research paper on AI investments from the Massachusetts Institute of Technology spooked the markets for tech stocks. Headlines underscored just one of the researchers’ findings: 95% of firms were seeing no return on their hefty AI investments. The more telling data got at why. Few companies have successfully integrated AI into their core business functions, despite the rollout of increasingly powerful models.
At RAND, our research on the macroeconomic implications of AI also found that adoption of generative AI into business practices is slow-going. By looking at recent census surveys of businesses, we found the level of AI use also varies widely by sector. For large sectors like transportation and warehousing, AI adoption hovered just above 2%. For finance and insurance, it was roughly 10%. Even in information technology—perhaps the most likely spot for generative AI to leave its mark—only 25% of businesses were using generative AI to produce goods and services.
Far from producing mass layoffs, AI appears to be—at least so far—increasing employment. Again using government census data, we found that while some businesses have decreased employment by using AI to replace tasks done by workers, a larger share has reported increases in employment related to AI adoption.
The disconnect between this data and the fears of a labor market apocalypse may stem, in part, from a conflation of a job’s exposure to AI with total automation of work. The worry is that if AI can perform a lot of tasks you do, maybe it can perform your whole job. That fear may be unfounded.
In a widely-referenced study, OpenAI estimated that 80% of the workforce has at least 10% of their tasks exposed to LLM-driven automation, and 19% of workers could have at least 50% of their tasks exposed. But jobs are more than individual tasks. They are a string of tasks assembled in a specific way. They involve emotional intelligence. Crude calculations of labor market exposure to AI have seemingly failed to account for the nuance of what jobs actually are, leading to an overstated risk of mass unemployment.
Our analysis did find that a substantial share of businesses plan to replace human tasks with AI in the near future. Whether this will lead to significant labor displacement, or just increased productivity, will depend on how many tasks AI will eventually be able to replace—but also how many new tasks AI will generate.
The pace of AI tech development is dizzying. For the larger economy, however, the rate of adoption matters most. Slow and uneven adoption isn’t necessarily negative: It means businesses are making deliberate choices on how best to integrate AI. That creates a window for workers to adapt, and for policymakers to manage what lies ahead.
About the authors: Jonathan Welburn is a senior researcher and Vegard M. Nygaard is an economist at RAND. They are both professors of policy analysis at the RAND School of Public Policy.
[ad_2]
Source link
