Welcome to “Eye on AI” with AI reporter Sharon Goldman. In this edition… New human studies on artificial intelligence and jobs… The leadership drama of Mira Murati’s thinking machines… Another blockbuster quarter for TSMC…Google Gemini introduces personal wisdom.
When people hear me cover AI, the first question they ask me is probably: “Will AI take over my job?” That question is often followed by: “How? your Work? “
sigh. Well, so far, I’m still here. New research from Anthropic shows that the fate of our jobs is more complex than the simple story of humans being replaced by artificial intelligence and robots.
In its latest issue economic indexAnthropic has launched a new way to measure how people actually use its chatbot Claude — what tasks they give it, how much autonomy they grant it, and how often it succeeds. The goal is to use data to get a clearer picture of whether AI can actually make people work faster, what types of tasks AI can best support, and how it might actually change people’s careers and the nature of their careers.
Yesterday, I spoke with Human Economist Peter McCrory about this ongoing research, and he said the company launched it in earnest a year ago, recognizing that AI is a general technology that will impact every job in some way—certainly every sector of the economy.
Artificial intelligence reshapes jobs in different ways based on different roles
But the results aren’t always so straightforward — at least not yet. For example, research has found that AI is reshaping jobs in different ways based on different roles. Radiologists or therapists may find that AI can enhance their skills by taking on some of their most time-consuming tasks, allowing them to spend more time talking to patients and clients. But people in other jobs may find that their skills are downgraded, or that their jobs are made simpler without allowing them to devote more time to some obviously higher-level tasks. This can happen with jobs such as data entry clerks, IT specialists, and travel agencies.
Additionally, human collaboration and oversight remain critical, especially for complex jobs. Therefore, AI appears to increase the productivity of highly skilled professionals rather than replace these roles.
McCrory said he hopes other researchers will use Anthropic’s insights to better understand AI’s uneven impact on the labor market. One thing is easy to understand, however: AI adoption is happening rapidly—in fact, AI is spreading faster in the United States than any major technology in the last century, he said.
“In our last report, released in September, we documented that use was disproportionately concentrated in a handful of states,” he explained. “In this report, we see evidence that low-use states are catching up quickly.”
The number of tasks Anthropic’s Claude can take on is constantly increasing
There’s no doubt that AI is getting more powerful: The number of tasks that Anthropic’s Claude can cover is growing, the study found, with 44% of jobs now able to use AI for at least a quarter of their tasks, up from 36% in the last report.
I couldn’t help but notice that the latest research was done before the debut of Anthropic’s latest model, Opus 4.5, and before the release of Claude’s Cowork app, a general-purpose artificial intelligence agent that can manipulate, read, and analyze files on a user’s computer, which was released just this week.
“It just goes to show how broadly applicable this technology is, and that Cloud is increasingly able to not only provide you with information, but to act on your behalf at your discretion and authority,” McCrory said. “I think you might see an increase in the importance of delegation skills—there’s evidence in the academic literature that people get more value out of large language models when they have better management skills.” He adds that this is his own experience: “I find myself delegating increasingly complex tasks to Claude that I might have given him had I had a research assistant.”
When I told McCrory I had written Last week’s “Focus on Artificial Intelligence” About a software developer who was excited about using Claude Code but was frustrated by having his role relegated to that of a manager, he nodded sympathetically and suggested I check it out Other human studiesdeveloped by its social impact team.
“I think the biggest takeaway here is we don’t know what’s going to happen,” he said. When I pointed out how much most of us dislike uncertainty, he emphasized that he hoped this report and data would help researchers see the future more clearly. “We’re committed to making this data open source,” he said, so economists and policymakers can better understand the potential of what’s coming and how we can prepare for it.
More AI news below.
Sharon Goldman
sharon.goldman@fortune.com
@sharongoldman
The wealth of artificial intelligence
Trump triggers retail investor selloff of Big Seven ——Jim Edwards
What the Apple-Google AI deal means for the two tech giants and upstart OpenAI worth $500 billion – Author: Jeremy Kahn and Beatrice Nolan
Artificial Intelligence News
The drama of leadership in Meera Murati’s The Thinking Machine. In a surprising leadership change, Barret Zoph and Luke Metz, two co-founders of Mira Murati’s artificial intelligence startup Thinking Machines Lab, have announced they are leaving the fledgling company to rejoin OpenAI, just months after leaving the organization to help launch the venture. According to “Wired”As part of the move, another former OpenAI researcher, Sam Schoenholz, will also return to OpenAI. The departures were confirmed by an internal memo from OpenAI application lead Fidji Simo, who said the return “has been in the works for weeks.” The turn of events is a major blow to Thinking Machines Lab, which only recently raised a large seed round and recruited top talent, and highlights the fierce competition for elite artificial intelligence researchers in the industry.
TSMC had another blockbuster quarter. According to CNBC TSMC, the world’s largest semiconductor maker, said profits rose 35% and revenue hit a record high as demand for artificial intelligence chips continues to surge. The company’s revenue and net profit beat expectations, and its high-performance computing business, driven by artificial intelligence and data center chips, now accounts for 55% of sales. Advanced chips of 7 nanometers or smaller account for more than three-quarters of wafer revenue, highlighting how cutting-edge artificial intelligence processors have become the core of TSMC’s business.
Google Gemini launches personal intelligence. in a New blog post Written by Josh Woodward, Google Vice President The company has announced the launch of a “personal intelligence” beta in its Gemini app, giving users the option to securely connect their Gmail, Photos, Search, YouTube and other Google apps to Gemini. The idea is to make the assistant more proactive and helpful — combining information from email, photos, and searches to answer questions or provide advice about your life — while keeping privacy control in the hands of the user (the feature is turned off by default, and users choose to connect content). Personal Intelligence is initially available in the U.S. to paid subscribers, with plans to expand over time, signaling that Google is leveraging its broader ecosystem to drive more personalized AI interactions to drive differentiation in its consumer AI.
Follow the AI Numbers
48%
How many single adults have reportedly used artificial intelligence to help draft breakup messages or texts that set boundaries? new research From chat assistant use.ai.
So now that users can rely on artificial intelligence to guide emotionally charged conversations, there’s no need for “ghost” dating partners anymore. among Of those who used an AI tool to gently drop a date, 62% said the resulting conversations were more structured and 39% reported less subsequent conflict.
According to the survey 4,812 single adults Across the five English-speaking markets, this trend isn’t limited to dating: 27% use AI to rehearse sensitive face-to-face conversations, while 20% use it to manage boundaries in non-romantic relationships.
you have a calendar
January 19-23: World Economic Forum, Davos, Switzerland.
January 20-27: AAAI Conference on Artificial Intelligence, Singapore.
February 10-11: Artificial Intelligence Action Summit, New Delhi, India.
March 2-5: Mobile World Congress, Barcelona, Spain.
March 16-19: Nvidia GTC, San Jose, CA
April 6-9: HumanX, San Francisco.