Is AI hurting American jobs? A bit, according to Federal Reserve Chair Jay Powell, whose comments Wednesday gave the AI fearmongers a good amount of material to work with. In a press conference, Powell said AI was “part of the story” behind the worsening unemployment that prompted the Fed to lower interest rates for the second time in two months.
Hi! If you’re finding value in our Applied AI newsletter, I encourage you to consider subscribing to The Information. It contains exclusive reporting on the most important stories in tech. Save up to $250 on your first year of access.
Welcome back!
Is AI hurting American jobs?
A bit, according to Federal Reserve Chair Jay Powell, whose comments Wednesday gave the AI fearmongers a good amount of material to work with.
In a press conference, Powell said AI was “part of the story” behind the worsening unemployment that prompted the Fed to lower interest rates for the second time in two months.
But he was quick to also say it was “not a big part of the story yet and we don’t know if it will be.”
Even though some companies say they are cutting staff because AI is handling more tasks (looking at you, Salesforce!) or saying they will hire fewer workers down the road (including Amazon, JPMorgan Chase and many others), such cuts aren’t showing up in broader economic indicators, he said. Unemployment insurance claims aren’t going up very much, for instance, he said.
The disconnect between corporate rhetoric on AI and jobless claims is “a little curious,” Powell said.
Perhaps Powell should talk to computer science professors, some of whom say graduates are having a hard time finding work! There are plenty of anecdotes, as well as legitimate studies, about young people straining to find good entry-level jobs in professions exposed to AI. Such examples might not be showing up in jobless claims figures yet.
The struggles of young software engineers make sense, given that coding is one of the top applications—if not the top one—for generative AI. Other potentially vulnerable fields include analytics, marketing and customer service, according to the handy chart in this article.
Powell is right to say it’s not a disaster yet. And it’s clear from our reporting that the generative AI revolution is taking longer to eliminate jobs or hiring in corporate America than many of us initially expected, even as new AI productivity tools become a regular part of software budgets. (Powell said he has been impressed by persistent productivity gains among American firms over the past five to six years, predating generative AI and likely spurred by pandemic-era automation efforts.)
Powell may not have certainty about the long-term effect of AI on the labor market, but he acknowledged that the current AI wave may be worse for labor than prior tech waves of the last 200 years. Those waves destroyed some jobs while creating new ones to the point where “you get new jobs and there are enough jobs” to go around.
The generative AI era “may be different,” he said. And if it is, “we don’t have the tools to deal with” the “social labor market implications.” Yikes!
Here’s what else is going on…
Bargain-Basement AI
The year 2026 could be the year of cheap AI.
That’s the tenor of comments from a growing number of corporate chief executives. They’ve been taking to the airwaves to note that free open-source or open-weight AI models—whose main selling point is that they cost less to run compared to proprietary AI from OpenAI and Anthropic—are good enough to power shopping assistants and customer service.
The latest to join the cheaper-is-better club is Pinterest CEO Bill Ready. The executive, who took the helm of the digital scrapbooking site three years ago after stints at Google and Paypal, said the company has been able to customize or fine-tune open-source models and get similar performance to that of the best proprietary models at less than 10% of the cost.
“Open source is at the table, and it is competing and doing so at really effective cost-to-performance levels,” he told Akash Pasricha on TITV, The Information’s streaming show, on Wednesday.
He highlighted DeepSeek’s models and Alibaba’s Qwen open-weight models, which Pinterest is testing. Pinterest also develops its own models for visual images, it outlined earlier this month. It’s part of an AI strategy designed to make sure Pinterest keeps up with chatbot-powered rivals while avoiding overwhelming the app with AI-generated “slop.”
Ready’s comments echoed those of Airbnb CEO Brian Chesky, who in October said Airbnb is “relying a lot” on Qwen to power an AI customer service agent. While Airbnb also uses OpenAI’s newest models for the same agent, Chesky said that “there are faster and cheaper models” that are a better fit for handling many queries.
Ready said he’s heard similar comments from many CEOs. “They’ve invested a lot of money in AI and buying these off-the-shelf proprietary software solutions, but they’re not seeing the return on investment that they need.”
It may not help that the prices of the most advanced proprietary models have stayed stubbornly high, as we reported this summer.
“Companies [are] finding that they can leverage open source to get much better efficiency,” he said.
We imagine he’s referring to sophisticated tech companies like Pinterest that know what they’re doing. The question is how many others have the expertise necessary to utilize open-source AI.—Laura Mandaro
