Redefining productivity in legal services
Generative AI doesn't have to cause mass unemployment or mass overwork in the legal profession. We can use it to free lawyers from task performance and enter a new era of client-centred productivity.
As I like to say, I’m not usually wrong, but I’m frequently too early. In a 2019 preface to a book on legal innovation, I wrote the following:
Think of all the amazing technologies that have arrived in the law over the last couple of decades, from document automation to contract drafting to e-discovery. Have lawyers enjoyed a windfall of unallocated hours and clear horizons with which to better ourselves and those around us, or engage in more fulfilling and higher-value endeavours?
Or have those “freed-up” hours been immediately captured by others and filled with ever more work, all in service of “greater productivity”? Especially if you work for an employer who measures your productivity in hours billed, and for whom “freed-up time for you” is the last thing they want?
Machine learning and artificial intelligence are only going to amplify and accelerate these trends and concerns.
And here we are, in 2025, when virtually every article about Generative AI in legal services says it will “free up” lawyers’ time for “higher-level activities.” Where will the benefits of this reallocation of lawyers’ time and effort likely accrue? To the lives of law firm labourers, or to the profits of law firm owners? You don’t have to be Karl Marx to correctly guess the answer.
Two recent articles drive this point home, and more. In “Jevons Paradox: A personal perspective,” Tina He observed that “productivity doesn’t eliminate work; it transforms it, multiplies it, elevates its complexity. … When everyone can produce 10x more, the baseline resets, leaving us all running faster just to stay in place.” And in “The Hyperproductivity Trap,” Tom Martin of LawDroid Manifesto noted:
If a research memo used to take you ten hours and now you can generate a half-decent first draft in two, you might find that your partner (or client, or judge) starts expecting two other projects to fill the “freed” eight hours. “With AI, that should be simple, right?” Or 80% of that time might be repurposed to refine and iterate on the AI’s output, making the end product more polished, but also piling on more versions and greater complexity. Instead of a breather, you get a busier day, with an even higher bar for quality.
It’s still possible that AI takes over so many legal tasks that the legal profession experiences mass unemployment. But it’s also possible we could go the other way, towards mass overwork.
In this scenario, AI eliminates simplicity in legal work (because it can do simple tasks in seconds) while also magnifying complexity (because its creative and analytical abilities reveal many more potential problems and opportunities to be addressed) — but it also intensifies volume (there’s just as much work or more for lawyers, but it’s all mission-critical, intricate, and highly concentrated).
Rather than benefitting from AI’s capacity to perform legal tasks, then, lawyers would be punished for it. Law firms would crack the whip on their “freed-up” lawyers, giving them even more work that further drains their intellectual and emotional reserves. Once again, the dividends from a technology advance would bypass the workers and drop straight to the owners’ bottom line.
That’s a pretty miserific vision. But there’s no reason to surrender to it, or regard it as inevitable. This future will manifest itself only if law firms insist, in the face of staggering technological change, on maintaining their old ways of working.
AI is not really the problem here. The problem is that many law firms don’t know how to do business except by keeping their lawyers as busy as possible with billable work. That model, whatever its merits and demerits, is just not compatible with a technology that compresses, intensifies, and upgrades the performance, volume, and quality of legal work. Something has to give. We have to break this deadlock.
And there is a way to do it — but it’s not going to be easy. AI is extending to the legal profession an offer we can’t afford to refuse: to finally redefine what “productivity” means in our industry.
Despite the fact — indeed, as evidenced by the fact — of lawyer overwork and burnout, the legal industry is actually in dire need of productivity enhancement. Most law firms are (in operational terms, anyway) brute-force, volume-based businesses. They confuse “effort” for “productivity”: The harder you work, they believe, the more productive you are.
Every other industry knows better. Actual productivity in the real world rises when outputs are increased (in terms of volume, quality, benefits, etc.), while inputs are decreased (in terms of effort expended, resources consumed, waste incurred, etc.).
The most effective way to increase productivity is through technology. Automation shifts basic low-value activity from people to machines, so that we can do or make more things, faster and to a higher degree of consistency and quality. Process improvements pave the way for automation and make it more effective once it’s implemented.
Law has resisted these trends more successfully than almost any other sector. Most legal services are still developed and delivered through individual lawyer effort, assembled piece by piece, hour by hour, in a painstaking linear process. High production costs not only render legal services scarce and expensive, but they also drive overworked human producers to burnout and depression. Law firms have adopted some efficiency upgrades over the years (e.g., document automation, e-discovery, project management), but they’ve been too incremental to bring about systemic change.
This will soon come to an end. AI, when fully applied to legal services, will produce efficiency gains so large as to nullify the legal industry’s one-dimensional, lawyer-centric definition of productivity. Much legal work will be performed with far less lawyer time and effort than before. (The question of AI’s resource consumption is a separate and important one that I’ll tackle in a future article.)
This doesn’t mean AI will eliminate human involvement in legal services delivery, of course. AI’s output will require lawyer oversight and verification for awhile yet; internal management of input resources (including the proficiency of lawyers) will require attention; and lawyers will always have to identify and monitor progress towards client goals.
But the important point is that lawyers are going to leave most legal task performance to the machines. They’re not going to be nearly as involved in the job-performing labour as they once were. Just as human diggers gave way to excavators and became site supervisors, or cargo handlers were replaced by warehouse robots and moved up to distribution logistics, lawyers will also ascend the value chain in legal services, where they can have greater impact on core client interests.
Constantly pushing lawyers to work harder will no longer be the way to law firm prosperity. Basing productivity on lawyer task performance won’t work anymore, because lawyers mostly won’t be doing “tasks” — the machines will. Lawyers will instead be focused on achieving client outcomes. As inputs (production time and resources consumed) fall, and outputs (more challenging problems and high-value solutions) rise, actual “productivity” will finally come to the legal sector.
Does this sound good to you? It does to me. But I need to be clear: Shifting from volume-based, input-first productivity to results-based, output-first productivity will be extremely challenging for the legal industry. Switching productivity definitions constitutes nothing less than a business-model transformation. Among the changes required:
Pricing systems based on lawyer effort will falter, as the speed of task performance vastly accelerates. While clients might be perfectly happy to stick with hourly billing once hours start nosediving, lawyers should work with clients to develop fees based on “output measures” such as objectives achieved, speed of performance, consistency of communication, and so forth (to the extent local ethics rules around billing allow). This one step alone would have a seismic impact on lawyer behaviour and law firm culture — a positive impact, but a seismic one all the same.
Compensation systems will be the next domino to fall after pricing. When a lawyer’s hour of attention to a client matter doesn’t automatically generate revenue for the firm, there’ll be no reason to reward (or record to the last minute) more such hours. What will bring money in the law firm door? New business and better results for new and existing clients — compensate that. What will improve efficiency and therefore increase profitability? Leaner operations, more automation, better lawyer training — compensate that too.
Lawyer development, as I’ve written about repeatedly, will undergo a fundamental shift. Associates will lose their utility as “billing drones,” but they can contribute greater value to client outcomes (and land their own new clients) if firms invest seriously in their accelerated development, mentoring, and supervised experiential training. The untapped potential of all these brilliant young lawyers will be evident once they finally emerge from the billable mines.
Data infrastructure will suddenly matter a lot more to firms. Your Generative AI systems will only be as good (and as competitively distinctive) as the data upon which they’re based, especially your firm’s homegrown legal knowledge banks. And if you want to demonstrate the effectiveness of the outcomes you’re generating for clients, you’ll need better metrics, dashboards, and other tools for tracking and establishing the quality of your solutions.
That’s a whole lot of disruption, in what will likely be a relatively short period of time (even assuming AI’s growing pains continue for the next few years and legal industry resistance creates enough friction to slow the process). But the upside!
You’d no longer have a business reason to keep a running tally of hours spent on a client engagement. Sure, you’d still need to monitor your time for internal purposes, but because the client isn’t paying you for your time, you won’t need to constantly worry, “Will the client pay for this?” and “Am I billing enough hours?”
You’d enter the happy kingdom of outcome-based pricing. No more hourly billing, no more discounting or writing off hours at the source, just a base fee that could vary (ethical rules permitting) according to success factors such as speed, absence of hassle, responsiveness, overall experience, and so forth.
You won’t be a volume labourer anymore; you’ll be a high-value legal service professional engaged in complex, challenging work. This will require, on top of your legal skills, time, patience, consultation, collaboration, conversation, and reflection. You’ll be an advocate, an advisor, and a steward of your client’s affairs.
That last point is especially important, and it’s the key to avoiding the overwork trap. Your client engagements will be sophisticated, ongoing, and relationship-based; they will require your intellectual and emotional commitment. You can’t take on too many of these, or the quality of your attention and service will suffer. And you won’t want to compromise the quality of this work, because it will be the most complex, challenging, fascinating, and fulfilling work you’ve ever been involved with.
It’s not going to happen tomorrow, or even next year. Maybe by the end of the decade, the shift will be well underway. But folks, this is where we’re headed. AI is going to cure our profession’s addiction to task-based lawyer productivity, whether we like it or not. The sooner your own law practice recognizes the inevitability of this change and begins making adjustments to accommodate it, the likelier you’ll enjoy the first and best fruits of the legal productivity revolution now in its earliest stages.
I think this is the more likely scenario. Just like with email, we were promised a faster, more efficient way to communicate. No longer would we have to write or type up a letter, address it, get postage, send it out, and then wait several days for the recipient to receive it, before having to wait a similar length of time for them to respond. With email you can send a message to someone and it will arrive within minutes. All that saved time could be used for more important things.
Except what happened is that we now spend an inordinate amount of our day managing our email — more time than we used to spend on written correspondence!
Same thing that happens with roads. When a road starts to get busy and traffic backs up, the typical solution is to build more lanes. In other words, increase the capacity. That helps for a time, but eventually more traffic uses the expanded road until more traffic uses it than did before. The problem just comes back bigger.
Work will expand to fill the space available. So if you create space by using AI to do work faster, all that will happen is that you will find more work (and as suggested, likely more complex) to fill the time.
Hi - how do you get around the vexatious litigant issue when it comes to value billing. As a prosecutor and defence counsel for various regulated professions, I find time and complexity intertwined with the approach of counsel on the other side. I have seen straightforward matters become behemoths.
When you discuss value billing and changing what productivity looks like to clients, how do you account for vexatious litigants? I am at jglick@gfsllp.ca and would welcome a conversation as I’d like to move in this direction but have gotten killed on flat fees and other arrangements for this issue. Much easier to value billing for advisory and other types of work that I do.