Behind the Curtain: AI, Automation, and the Looming Threat to Human Knowledge Work

by

Inspired by Axios’s “Behind the Curtain: A White-Collar Bloodbath” (May 28, 2025)

Dario Amodei, cofounder and CEO of Anthropic, is issuing an urgent warning: advanced artificial intelligence may soon pose a serious threat to millions of white-collar jobs. While today’s AI systems, like Anthropic’s own Claude and OpenAI’s ChatGPT, are currently seen as productivity boosters, Amodei cautions that this could quickly change as models become dramatically more powerful.

In internal presentations recently shared with government officials, Amodei projected that future AI models, potentially arriving in the early 2030s, could be capable of performing 80 to 90% of tasks typically handled by college educated professionals. These include jobs in legal research, finance, marketing, and customer service. For example, AI tools are already being deployed to automate paralegal tasks and financial analysis; and some early adopter companies are replacing portions of their human customer support teams with large language model (LLM) chatbots.

These concerns aren’t just hypothetical. Anthropic’s own forecasting suggests that without meaningful intervention, future models could lead to what the company has called a “white-collar bloodbath,” reshaping employment in the professional services sector on a scale not seen since the Industrial Revolution.

One field facing both disruption and opportunity is law librarianship. Traditionally seen as the gatekeepers of legal information, law librarians are already witnessing a shift as AI tools increasingly take on roles in legal research, document summarization, and information retrieval. Systems powered by LLMs can scan and analyze large volumes of case law, statutes, and regulations in seconds, tasks that once took hours of careful human curation. While this may increase efficiency, it also raises concerns about the devaluation of specialized human judgment and the long term stability of information-based roles.

However, many in the legal information community see AI not as a replacement but as an enhancement to the profession. Law librarians can serve as critical intermediaries who evaluate, implement, and oversee AI systems to ensure accuracy, mitigate bias, and protect against overreliance. As these technologies evolve, the role of the law librarian may shift from “finder of facts” to “manager of machines”, shaping ethical research frameworks, training users, and preserving human oversight in a rapidly automating legal ecosystem.

Amadi is hardly hopeless. He sees a variety of ways to help prevent the worst case employment scenarios, and has proposed a forward-looking policy idea: a “token tax.” This would place a small levy on each digital token, or unit of language, processed by powerful AI models. Since large scale deployments of AI involve billions of tokens per day, even a modest tax could raise significant revenue. The funds could then be used to finance job retraining programs, education initiatives, or broader social safety nets such as universal basic income.

Reuven S. Avi Yonah, Professor of Law and Director of the international tax LLM  Program at the University of Michigan has concerns that align closely with those  of Amodei.  He believes that incentives created through taxation could play invaluable roles in limiting “hallucinations,” copyright infringement, and  other  problems connected to AI.  For a discussion of his proposals in detail see his article,  Can Tax Policy Help Us Control Artificial intelligence, in the Summer 2025 issue of the Marquette Lawyer. He would like to see a targeted policy  that taxes only the AI program and not the corporation per se.

“We need to think now about how to manage a world where AI can do a majority of economically valuable tasks,” Amodei said. “The disruption will be profound if we’re not prepared.”

Amodei’s message aligns with growing calls from within the tech sector for anticipatory regulation. But so far, few concrete policies have emerged at the federal level. As companies race to develop ever more capable AI systems, the clock is ticking on the opportunity to shape a humane and equitable transition for law librarians and many others navigating the shifting landscape of white collar work.

Further Reading:

Contact Information
ATTORNEY ADVERTISING

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Please do not include any confidential or sensitive information in a contact form, text message, or voicemail. The contact form sends information by non-encrypted email, which is not secure. Submitting a contact form, sending a text message, making a phone call, or leaving a voicemail does not create an attorney-client relationship.

Copyright © 2016 – 2025, David Badertscher
JUSTIA Law Firm Blog Design