As AI grows more capable by the day, headlines warning of mass job displacement have become impossible to ignore. But early evidence suggests the reality is far more complicated.
Snowflake recently published a report that found AI-driven job creation is outpacing job losses. In a survey of 2,050 business and technology leaders across 10 countries, far more reported AI creating jobs than eliminating them: 11% said AI had cut roles at their organizations, 42% said it had created new ones, and another 35% reported a mix of both.
“While automation will continue to advance, what we’re observing in the data is less about outsourcing entire roles and more about reshaping them to support AI-driven workflows,” said Anahita Tafvizi, Chief Data Analytics Officer at Snowflake, to The Deep View.
For instance, IT operations saw a 56% job gain but also a 40% job loss, reflecting that AI can significantly impact workflows: it creates opportunities for expansion while also enabling complete automation.
Among the other jobs showing gains were cybersecurity and software development, fields where AI can perform the work proficiently and that rank among the highest in AI-observed exposure, according to a new Anthropic report.
Anthropic developed this new measure called observed exposure, which combines theoretical LLM capabilities with real-world usage data to more accurately assess AI job displacement.
Occupations with higher observed exposure were concentrated in white-collar roles such as legal, computer and math, finance, arts and media, education and library, and are projected by the BLS to grow less through 2034.
Yet, there has been no systematic increase in unemployment for highly exposed workers since late 2022. At most, the report found that hiring of younger workers has slowed in exposed occupations, likely signaling that companies are slowing hiring of inexperienced workers, as AI can likely do their roles more cheaply and faster.
“From a practical standpoint, the most important action is investing in adaptability,” said Tafvizi. “That doesn’t necessarily mean becoming an AI engineer. It means building data fluency, strengthening AI literacy, and deepening domain expertise so you can work effectively alongside these systems.”
Our Deeper View
While it may feel like AI news is everywhere you look, it is important to remember that the technology only began gaining momentum at the start of 2023, so it is still relatively new. It is therefore far too soon to determine what lasting impact it will have on jobs. Although company leaders have attributed layoffs to AI-related causes, AI often serves as a convenient scapegoat for broader cost-cutting decisions. This isn't to say that some jobs aren't being offloaded to AI, but be wary of sweeping statements. Company leaders often face pressure from shareholders to improve their financials, and reducing headcount is one of the quickest ways to make the numbers look better. What better cause to blame than an emerging technology that people already fear, and that makes companies appear to be on the cutting edge?