The Financial Times just published an article titled “Wall Street banks push Big Law to cut fees because of AI.” Apparently the investment bankers have decided that paying legal bills doesn’t make sense now that ChatGPT can spit out answers. We’re mere months removed from the high profile — yet very entertaining — disaster of a gaming CEO asking ChatGPT to give him the legal basis to avoid paying a $250 million bonus.
Yadda yadda yadda, he lost.
Has anything happened in the intervening months to make financial institutions trust turning over their legal future to AI? No. But they’re asking for it anyway. The next recession is going to be wild, y’all.
Just last week, at ILTACON, the message from clients was that they wanted outcomes at any price. Now Wall Street, the people who brought you the Great Recession, wants its law firms to turn over legal review to the robots.
This is funny, but there’s actually a lot of truth here.
The monied interests shoveling cash onto the AI bonfire, clucking their tongues knowingly that OF COURSE an industry that’s burned almost a trillion and a half in expenses to generate around $200 billion is a sound investment! The only way that bet works is if you accept the deeply misanthropic worldview of Silicon Valley investors who believe AI will become a digital messiah — a robot Jesus, except with less “healing the sick,” and more “shoving poor people into the human battery farms from the Matrix.” For these people, AI makes sense because it can replace human labor at every level (except, for some reason, their own ability to invest).
The finance bros believe the robot is infallible because it’s a necessary article of faith when they’ve loaded up their books with a money black hole. They don’t have time to worry themselves with hallucinations. Or boilerplate drawn from wholly unrelated deals. Or suboptimal research. Just press the “Easy” button and turn in whatever the hell comes out!
But that’s not how law works. Without human legal judgment, mistakes get made. After a couple years of “human-in-the-loop” — an empty phrase that presumes a senior partner will read 200 pages of agentic logs on the backend — legal tech is starting to realize that the process needs to optimized to give the AI more break points for humans to interject, course correct, and reflect.
AI keeps getting faster, but the human brain doesn’t. Those hours and days spent turning the document included epiphanies and strategic rethinks. Compressing the time from start to finish robs humans of that time to really think about whether they’ve forged the right output. This matters because AI delivers the median acceptable answer by design. It’s taking prior results and spitting out smoothed over mediocrity. That may give a lawyer a good head start, but it’s not an endpoint. And bringing the output up to snuff requires more than a once-over with a red pen. It’s going to require real time.
At ILTACON, one analogy I made after the panel I spoke on is that the investors want AI to be a robot, but the law needs it to produce a cyborg. From this industry’s perspective it has to work in conjunction with humans. Indeed, we’ve already seen some early indications that AI isn’t reducing lawyer hours but increasing them. AI can surface research rabbit holes that humans might never have found before, and the humans are chasing them down to deliver better work product.
Not that AI won’t change billing. To the extent it saves time in some places, that is time that lawyers can’t bill. In the past, people mused about the $10,000 billable hour to reflect the value of senior judgment. The banks, it seems, aren’t buying it. A flat fee for relatively repeatable tasks could give the client cost predictability without firms sacrificing revenue. That makes the most sense, but it requires clients to believe they’re getting value.
If this is the stance they want to take, they’ll discover the value in a year or so when the litigation gets going. Because they’re playing a risky game with bet-the-company propositions. Of course, considering that they’re continuing to hand hundreds of billions of dollars to OpenAI in 2026, their betting judgment may leave a lot to be desired anyway.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.