A concise, curated roundup of recent AI, emerging technology, and cyber developments with implications for state attorney general offices. In this issue, I focus on AI’s potential impact on the practice of law within the private bar, as well as the issue of generative AI sycophancy that I noted in Issue #1. These will be recurring themes to which I will return in future editions.

Issue 1: AI is transforming legal work whether legal organizations prepare for it or not.

Jordan Furlong is a legal futurist and consultant to legal organizations across Canada and the United States. When law firms need advice on how to strategically position themselves amid emerging trends in the legal market, they call Jordan. So when he addressed this year’s American Bar Association Tech Show in Chicago, attendees paid close attention. I was one of those attendees.

Jordan set out to answer two questions in his keynote remarks:

As AI displaces lawyers from a growing share of legal task performance, what will be left for lawyers to do? And how are we going to develop lawyers when we don’t know what we’re training them for, and when we can’t count on law firms to do the training anymore?

According to Jordan, lawyers—especially those in law firms—will no longer be indispensable in a world where generative AI mechanizes legal production. AI will supplant lawyers as the primary engine of legal work, driving massive productivity gains.

AI operates faster and ultimately at lower cost, displacing the billable hour. Humans are not designed to scale; AI is—and in the legal domain, it is improving rapidly.

For the average person, the scarcity of elite legal knowledge—once accessible only to those who could afford exorbitant hourly rates—may come to an end. As the fundamentals of the profession shift, law firms will undergo significant consolidation and a shrinking of their traditional business model.

This will displace some mid- and long-tenured attorneys and, ultimately, eliminate the first-year associate role as the firm-as-training-ground model collapses.

What becomes of the attorney in this post-AI world? According to Jordan Furlong, that world is already here—which means we shouldn’t look ahead so much as look around. To answer the question, he argues, we must start by building a new competence profile for the modern lawyer.

The post-AI lawyer will no longer be hired for what they know, but for how they judge, guide, and lead. In other words, AI handles the legal tasks; attorneys handle the human consequences by employing judgment, exercising discretion, and ultimately providing guidance to clients.

Jordan provides a compelling and impassioned—if also unsettling—view of a world that has crept up on us. Whether one accepts Jordan’s interpretation is another matter, but certain things seem incontrovertible:

Here are a few quick takeaways:

  • The legal profession is in the midst of a profound change in how we operate. What that looks like, and how it impacts actual jobs (especially entry-level jobs), remains to be seen;
  • Most law schools are behind in adapting their pedagogical approach, and that chasm could become even wider; and
  • Current attorneys—especially those early- to mid-career—need to adapt quickly, or they risk being left behind.

Further Reading

Issue 2: A law firm powered by generative AI

Building on Jordan Furlong’s keynote address, there is an interesting article on X worth your time. Zack Shapiro is the founding partner of Rains LLC, a startup boutique firm that, according to Shapiro, competes with global firms in “startup formation, venture capital transactions, and regulatory work.” In an article entitled “The Claude-Native Law Firm,” Shapiro details how he uses generative AI—specifically Anthropic’s Claude—to compete in this marketplace.

Using the example of a client acquisition closing that included a last-minute request to restructure certain aspects of the deal, Shapiro explains that his AI-enabled approach not only allows him to compete, but that it moves faster, produces more thorough work product, and operates at a cost structure that would have been impossible 18 months ago.

The article explains how Shapiro develops workflows using Claude and serves as a compelling illustration of Jordan Furlong’s vision of the future of law. Other analysts believe that what Shapiro describes is an outlier and does not threaten either large law firms or small- and medium-sized firms. Nonetheless, it remains a fascinating read and may be instructive for state attorney general offices seeking concrete use cases, effective workflow integration, and guidance on training staff to use generative AI responsibly and effectively.

Further Reading

Issue 3: Stanford research demonstrates that AI sycophancy is a feature and not a bug

Researchers at Stanford tested 11 leading AI chatbots and found that AI sycophancy is not only prevalent but also associated with harmful outcomes. Across 11 AI models, AI affirmed users’ actions substantially more often than humans on average, including in cases involving deception, illegality, or other harms.

Lead author Myra Cheng told The New York Times that the most striking—and troubling—finding was the extent to which this behavior distorted users’ attitudes and judgments. Despite this, she noted, people tend to trust and favor a chatbot’s output, further driving engagement and creating perverse incentives to sustain—or even amplify—the behavior. Experts such as public-interest technologist and intellectual Bruce Schneier have argued that this kind of sycophancy is ultimately a design decision.

Further Reading

Author

Related Posts