{"id":167028,"date":"2026-11-09T14:02:07","date_gmt":"2026-11-09T22:02:07","guid":{"rendered":"https:\/\/xira.com\/p\/2026\/11\/09\/it-all-hinges-on-judgment-the-work-not-counted\/"},"modified":"2026-11-09T14:02:07","modified_gmt":"2026-11-09T22:02:07","slug":"it-all-hinges-on-judgment-the-work-not-counted","status":"publish","type":"post","link":"https:\/\/xira.com\/p\/2026\/11\/09\/it-all-hinges-on-judgment-the-work-not-counted\/","title":{"rendered":"It All Hinges On Judgment: The Work Not Counted"},"content":{"rendered":"
\"\"
<\/figcaption><\/figure>\n

I recently wrote about Torsten Slok\u2019s argument<\/a> that outside of big tech, AI has delivered zero financial payoff to corporate America. I suggested that could be even more true in legal. And I promised more on whether and how AI really reduces hours, an argument<\/a> Melissa Rogozinski and I have made before. So here goes.<\/p>\n

After reading Harshit Garg<\/a>\u2018s recent article in Legal Insider<\/a>, I wonder if the reason we aren\u2019t seeing the true financial payoff of AI in legal is not because AI doesn\u2019t save time but because we aren\u2019t good at counting all the time it takes to turn AI\u2019s output into something useful. Time that requires judgment that AI can\u2019t provide.<\/p>\n

Garg is the founder and CEO of a company that builds agentic AI for lawyers. He notes that AI companies measure the cost of generating output. They can tell customers how many tokens were consumed, how quickly a document was produced, and what the generation cost was.<\/p>\n

But then lawyers must review, verify, correct, revise, and ultimately take responsibility for that document. And that takes time, and time costs money. Garg\u2019s point is that AI usage and subsequent human work often aren\u2019t connected in a way that lets firms calculate the true cost of an AI-assisted task internally.<\/p>\n

The firm sees the hours AI eliminated. What it doesn\u2019t see is the extra review time AI created. Which means it overestimates what AI is costing it in billable hours. That puts management under a microscope with its partners who wonder why so much money is being spent on AI tools. So, management is prone to say: Why fool with AI? It\u2019s just costing us money and creating headaches.<\/p>\n

A Hypothetical<\/strong><\/p>\n

Let\u2019s say a lawyer traditionally spends four hours drafting a document and one hour reviewing it. With AI, the initial draft takes minutes, but the lawyer now spends two hours reviewing, checking, and correcting the output. Compare the two approaches:<\/p>\n

\n\n\n\n\n\n\n
<\/td>\nTraditional<\/strong><\/td>\nAI-assisted<\/strong><\/td>\n<\/tr>\n
Drafting<\/td>\n4 hours<\/td>\n0.1 hours<\/td>\n<\/tr>\n
Review<\/td>\n1 hour<\/td>\n2 hours<\/td>\n<\/tr>\n
Total<\/td>\n5 hours<\/td>\n2.1 hours<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n

AI has unquestionably saved significant time, some 60%. But that\u2019s a far cry from saving almost all of the time it would have taken without AI. The savings aren\u2019t nearly as dramatic as the reduction in drafting time suggests. And the loss of revenue isn\u2019t as dramatic as it appears either.<\/p>\n

And depending on who performs the review, the dollar savings could be even smaller. A partner reviewing AI-generated work will often cost substantially more than an associate preparing the initial draft. For example: the associate drafts at $250 an hour and the partner reviews at $800. Traditional: $1,000 + $800 = $1,800. With AI: $25 + $1,600 = $1,625. AI cut the hours by 60%. It cut the dollars by less than 10%.<\/p>\n

Which means the hidden review work may shrink both the promised savings and the revenue losses firms fear. The point being, the appropriate comparison isn\u2019t drafting time versus AI generation time. It\u2019s the total cost of producing an acceptable final product with and without AI.<\/p>\n

But What About The Client?<\/strong><\/p>\n

And there\u2019s another misunderstanding that creates unrealistic expectations. Clients have been hearing for years about AI\u2019s ability to reduce the cost of legal services, and they conclude AI can produce a document in seconds. End of story.<\/p>\n

But when the legal bill arrives, they naturally wonder: Where the hell are all my savings?<\/p>\n

The firm explains that the document still requires professional review, verification, and correction. That\u2019s a legitimate explanation. But the client may be skeptical. Particularly if the firm\u2019s marketing has emphasized dramatic AI-driven efficiency gains.<\/p>\n

And if the firm cannot connect the AI-generated work to the review time it required, and explain that in ways the client can understand, justifying the bill becomes even harder. The difference between promised savings and actual savings may create a credibility problem for law firms.<\/p>\n

I can tell you from experience when a client is expecting a 90% reduction in fees but only gets 30%, they ain\u2019t happy. And any explanation falls on deaf ears since they figure they\u2019ve been gouged.<\/p>\n

It\u2019s not simply whether the firm is charging appropriately. It\u2019s whether the client believes the firm is delivering the economic benefits of AI. That may explain why clients think they aren\u2019t getting any AI savings, as\u00a0I have previously discussed<\/a>. And why\u00a0Clio<\/a>\u00a0found that 57% of consumers (aka clients) think faster work should result in lower cost. Period.<\/p>\n

Which Brings Us Back To Slok<\/strong><\/p>\n

Slok found that outside big tech, AI\u2019s promised payoff hasn\u2019t shown up in the numbers. Garg offers one reason why. In essence, companies measure how fast AI does the task. They don\u2019t measure the human work it takes to finish it. Across corporate America, that gap could be swallowing a lot of the promised productivity.<\/p>\n

The Real Cost Of Getting To Done<\/strong><\/p>\n

But that gap has real impacts. AI may generate a document in seconds. But someone still has to make sure it\u2019s right. Someone has to check the facts, assess the legal reasoning, verify the citations, and take responsibility for the result. And that work has a cost. What matters isn\u2019t how much time AI saves on a first draft. It\u2019s how much it saves on a final product a lawyer can confidently hand to a client.<\/p>\n

Failing to understand that has consequences. Firm management and partners may conclude they are taking a financial beating on expensive AI tools. That can lead to strife within the firm. Or management dragging its feet. Clients, on the other hand, may have unrealistic expectations about how much AI will reduce their fees. And lose sight of what lawyers uniquely bring: judgment. That\u2019s a dangerous combination. And it could reduce the quality of the legal work.<\/p>\n

Until we measure and understand the value and cost of both AI and the proverbial human in the loop, all those claims about AI ROI may be little more than educated guesses. Guesses that lead to assumptions. Assumptions that could put firms and clients at odds. And firm management and partners at each other\u2019s throats. Neither is good.<\/p>\n

\n

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes\u00a0<\/em><\/strong>TechLaw Crossroads<\/em><\/strong><\/a>, a blog devoted to the examination of the tension between technology, the law, and the practice of law<\/em><\/strong>.\u00a0<\/p>\n

\n

The post It All Hinges On Judgment: The Work Not Counted<\/a> appeared first on Above the Law<\/a>.<\/p>\n

\"\"
<\/figcaption><\/figure>\n

I recently wrote about Torsten Slok\u2019s argument<\/a> that outside of big tech, AI has delivered zero financial payoff to corporate America. I suggested that could be even more true in legal. And I promised more on whether and how AI really reduces hours, an argument<\/a> Melissa Rogozinski and I have made before. So here goes.<\/p>\n

After reading Harshit Garg<\/a>\u2018s recent article in Legal Insider<\/a>, I wonder if the reason we aren\u2019t seeing the true financial payoff of AI in legal is not because AI doesn\u2019t save time but because we aren\u2019t good at counting all the time it takes to turn AI\u2019s output into something useful. Time that requires judgment that AI can\u2019t provide.<\/p>\n

Garg is the founder and CEO of a company that builds agentic AI for lawyers. He notes that AI companies measure the cost of generating output. They can tell customers how many tokens were consumed, how quickly a document was produced, and what the generation cost was.<\/p>\n

But then lawyers must review, verify, correct, revise, and ultimately take responsibility for that document. And that takes time, and time costs money. Garg\u2019s point is that AI usage and subsequent human work often aren\u2019t connected in a way that lets firms calculate the true cost of an AI-assisted task internally.<\/p>\n

The firm sees the hours AI eliminated. What it doesn\u2019t see is the extra review time AI created. Which means it overestimates what AI is costing it in billable hours. That puts management under a microscope with its partners who wonder why so much money is being spent on AI tools. So, management is prone to say: Why fool with AI? It\u2019s just costing us money and creating headaches.<\/p>\n

A Hypothetical<\/strong><\/p>\n

Let\u2019s say a lawyer traditionally spends four hours drafting a document and one hour reviewing it. With AI, the initial draft takes minutes, but the lawyer now spends two hours reviewing, checking, and correcting the output. Compare the two approaches:<\/p>\n

\n\n\n\n\n\n\n
Traditional<\/strong><\/td>\nAI-assisted<\/strong><\/td>\n<\/tr>\n
Drafting<\/td>\n4 hours<\/td>\n0.1 hours<\/td>\n<\/tr>\n
Review<\/td>\n1 hour<\/td>\n2 hours<\/td>\n<\/tr>\n
Total<\/td>\n5 hours<\/td>\n2.1 hours<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n

AI has unquestionably saved significant time, some 60%. But that\u2019s a far cry from saving almost all of the time it would have taken without AI. The savings aren\u2019t nearly as dramatic as the reduction in drafting time suggests. And the loss of revenue isn\u2019t as dramatic as it appears either.<\/p>\n

And depending on who performs the review, the dollar savings could be even smaller. A partner reviewing AI-generated work will often cost substantially more than an associate preparing the initial draft. For example: the associate drafts at $250 an hour and the partner reviews at $800. Traditional: $1,000 + $800 = $1,800. With AI: $25 + $1,600 = $1,625. AI cut the hours by 60%. It cut the dollars by less than 10%.<\/p>\n

Which means the hidden review work may shrink both the promised savings and the revenue losses firms fear. The point being, the appropriate comparison isn\u2019t drafting time versus AI generation time. It\u2019s the total cost of producing an acceptable final product with and without AI.<\/p>\n

But What About The Client?<\/strong><\/p>\n

And there\u2019s another misunderstanding that creates unrealistic expectations. Clients have been hearing for years about AI\u2019s ability to reduce the cost of legal services, and they conclude AI can produce a document in seconds. End of story.<\/p>\n

But when the legal bill arrives, they naturally wonder: Where the hell are all my savings?<\/p>\n

The firm explains that the document still requires professional review, verification, and correction. That\u2019s a legitimate explanation. But the client may be skeptical. Particularly if the firm\u2019s marketing has emphasized dramatic AI-driven efficiency gains.<\/p>\n

And if the firm cannot connect the AI-generated work to the review time it required, and explain that in ways the client can understand, justifying the bill becomes even harder. The difference between promised savings and actual savings may create a credibility problem for law firms.<\/p>\n

I can tell you from experience when a client is expecting a 90% reduction in fees but only gets 30%, they ain\u2019t happy. And any explanation falls on deaf ears since they figure they\u2019ve been gouged.<\/p>\n

It\u2019s not simply whether the firm is charging appropriately. It\u2019s whether the client believes the firm is delivering the economic benefits of AI. That may explain why clients think they aren\u2019t getting any AI savings, as\u00a0I have previously discussed<\/a>. And why\u00a0Clio<\/a>\u00a0found that 57% of consumers (aka clients) think faster work should result in lower cost. Period.<\/p>\n

Which Brings Us Back To Slok<\/strong><\/p>\n

Slok found that outside big tech, AI\u2019s promised payoff hasn\u2019t shown up in the numbers. Garg offers one reason why. In essence, companies measure how fast AI does the task. They don\u2019t measure the human work it takes to finish it. Across corporate America, that gap could be swallowing a lot of the promised productivity.<\/p>\n

The Real Cost Of Getting To Done<\/strong><\/p>\n

But that gap has real impacts. AI may generate a document in seconds. But someone still has to make sure it\u2019s right. Someone has to check the facts, assess the legal reasoning, verify the citations, and take responsibility for the result. And that work has a cost. What matters isn\u2019t how much time AI saves on a first draft. It\u2019s how much it saves on a final product a lawyer can confidently hand to a client.<\/p>\n

Failing to understand that has consequences. Firm management and partners may conclude they are taking a financial beating on expensive AI tools. That can lead to strife within the firm. Or management dragging its feet. Clients, on the other hand, may have unrealistic expectations about how much AI will reduce their fees. And lose sight of what lawyers uniquely bring: judgment. That\u2019s a dangerous combination. And it could reduce the quality of the legal work.<\/p>\n

Until we measure and understand the value and cost of both AI and the proverbial human in the loop, all those claims about AI ROI may be little more than educated guesses. Guesses that lead to assumptions. Assumptions that could put firms and clients at odds. And firm management and partners at each other\u2019s throats. Neither is good.<\/p>\n

\n

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes\u00a0<\/em><\/strong>TechLaw Crossroads<\/em><\/strong><\/a>, a blog devoted to the examination of the tension between technology, the law, and the practice of law<\/em><\/strong>.\u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"

I recently wrote about Torsten Slok\u2019s argument that outside of big tech, AI has delivered zero financial payoff to corporate America. I suggested that could be even more true in legal. And I promised more on whether and how AI really reduces hours, an argument Melissa Rogozinski and I have made before. So here goes. […]<\/p>\n","protected":false},"author":3,"featured_media":167020,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[16],"tags":[],"class_list":["post-167028","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-above_the_law"],"_links":{"self":[{"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/posts\/167028","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/comments?post=167028"}],"version-history":[{"count":0,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/posts\/167028\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/media\/167020"}],"wp:attachment":[{"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/media?parent=167028"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/categories?post=167028"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/xira.com\/p\/wp-json\/wp\/v2\/tags?post=167028"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}