
It’s a thriller. Most in-house counsel need their exterior companies to aggressively use AI, and most anticipate their payments to be diminished. But, they hardly ever demand these items from their exterior companies. It’s a counterintuitive hole. What offers?
A current Axiom study sheds some mild. Put merely, plenty of in-house counsel don’t have any extra of a clue the best way to use AI successfully than do their exterior companies. If the folks paying the payments don’t perceive what AI can realistically accomplish, how can they know what to demand? How can they maintain the folks sending the payments accountable?
The Research and Its Supply
Now we do want to start out with a little bit wholesome skepticism. Axiom is another authorized service supplier (ALSP). Its major mission is to provide in-house authorized departments with authorized expertise for particular and infrequently non permanent wants. So it’s in Axiom’s curiosity if in-house authorized wants assist with AI. That doesn’t essentially imply the survey outcomes are improper. I’ve no proof of that. Nevertheless it’s all the time good to bear in mind who’s paying for work like this.
Axiom sponsored the examine and commissioned InsightDynamo, which describes itself on its web site as a healthcare analysis agency (Hmm …). InsightDynamo surveyed 528 in-house authorized professionals on-line at corporations with annual revenues of not less than $100 million. The survey was finished between March 13, 2026, and March 25, 2026.
A Lack Of Metrics, Abilities, And Use
The outcomes paint a bleak image of in-house authorized’s understanding and efficient use of AI. Of these surveyed, 83% don’t know the best way to measure AI’s ROI or assess how their division’s AI spending is figuring out. Meaning fewer than 20% have established metrics to evaluate the effectiveness of what they’ve spent.
Solely 7% are optimizing and measuring AI throughout their group. Take into consideration that: they’re spending plenty of cash on AI however haven’t any clue if it’s producing significant returns.
Most respondents assessed their AI expertise as primarily intermediate. Solely 10% imagine that they’re superior. There’s a motive for that: solely 25% ship their personnel out for exterior coaching. Axiom notes that departments are growing AI spending “with a workforce that has had at most one working day of coaching.” Jeez.
Solely 17% of these surveyed report {that a} majority of their staff use AI instruments every day. In line with the report, just one in 4 use AI month-to-month. As Axiom places it, “the talents hole shouldn’t be entry or adoption; it’s coaching and common expertise utilizing the instruments.” Solely 40% have applied AI planning and governance practices. One other 40% say they plan to, which suggests they haven’t but. (Or as they are saying the place I grew up, they’re fixin’ to. Proper.)
The obstacles that the respondents determine to figuring out AI worth are additionally instructive. They embrace:
– a perception that the workflow adjustments can be too disruptive
– a sense that group members will resist change
– the view that expertise are missing
– a way that ROI is simply too arduous to measure
Sound acquainted? It’s the identical listing you’d most likely get from a room filled with legislation agency managing companions.
Nonetheless, 95% magically anticipate exterior counsel to make use of AI (and use it successfully), and 92% anticipate decrease charges when companies do use AI, an expectation that’s largely unhappy. However another survey shows that 80% of in-house don’t require and even encourage their companies to make use of AI.
Authorized AI? What Authorized AI?
And what AI instruments are getting used? 67% use general-purpose or frontier fashions, not legal-specific fashions. Instruments like ChatGPT, Copilot, and Gemini. The identical instruments most of us use for nonlegal work. Because the report places it, “Most authorized groups are working AI the identical manner a advertising and marketing division would.” That means most authorized departments don’t know what they’re doing with AI.
Choice Course of? What Choice Course of?
And the choice course of for AI instruments? In a phrase: abysmal. Only one in 10 groups conduct rigorous analysis earlier than figuring out what product to purchase. Fewer than 1 in 4 run structured pilots with any kind of outlined objectives, timelines, or measurement.
When you don’t know what you need a software to perform before you purchase it, it’s fairly arduous to find out whether or not it achieved something afterward. It’s no marvel the ROI is elusive.
Steerage? What Steerage?
And in-house groups are determined for steerage. Nearly all, 98%, say they need they’d exterior steerage on software picks, and 96% say that they’d use an ALSP for adoption or implementation recommendation, not a legislation agency. That is most likely as a result of ALSPs typically have an financial incentive to make use of AI and are due to this fact perceived as additional alongside the training curve. It additionally displays a scarcity of belief in legislation companies. (Axiom is an ALSP.)
What Does It All Imply?
It’s not stunning that authorized departments aren’t seeing their payments diminished. In-house authorized departments don’t actually know the best way to use AI successfully. The numbers clearly present an actual lack of understanding and understanding amongst in-house authorized groups with regards to AI. They’ll’t work out the return on funding, which suggests they will’t determine the place AI can be helpful or the best way to get there.
If they will’t work out the worth, how within the hell can they inform exterior counsel the best way to use the instruments in ways in which cut back their payments? How can they’ve a significant dialog with exterior counsel about the place AI needs to be used to scale back charges with out sacrificing high quality?
Furthermore, exterior counsel clearly gained’t use AI to scale back their payments till purchasers demand it. To begin with, most don’t have any extra of a clue about its use than in-house. Extra importantly, these companies that invoice by the hour (as do most of the exterior companies that work for these corporations) haven’t any financial incentive to make use of instruments and applied sciences that they worry will finally cut back their income. As Richard Susskind famously famous, it’s arduous to inform a room filled with millionaires their enterprise mannequin is all improper. It’s even tougher to persuade them.
In-house counsel, however, have the financial incentive to make use of AI. However they aren’t utilizing it as successfully as they need to. They don’t know which product to make use of, nor do they know the best way to conduct a strong analysis to resolve what to make use of. And similar to their exterior brethren, they view AI use as disruptive and resist change.
In order that they’re not a lot completely different from the skin companies they now anticipate to make use of AI regardless of these companies’ disincentive to take action. Maybe that expectation shouldn’t be unreasonable, since exterior companies are within the enterprise of serving their purchasers. So it might be of their purchasers’ greatest pursuits for them to make use of AI. That could be true for some companies, however that’s merely not actuality for many.
The Blind Main The Blind
So right here we sit in 2026. You’ve legislation companies reluctant to make use of AI and frightened of what it might imply for his or her economics. You’ve in-house authorized departments that don’t perceive AI’s worth or the best way to use it successfully. That creates a cussed hole between what in-house counsel need and what exterior companies are doing.
However until and till in-house authorized departments higher perceive the best way to use AI and what might be realistically demanded of outdoor companies, that hole will stay. You may’t demand what you don’t perceive.
Need your exterior companies to make use of AI successfully? Discover ways to use it in-house after which demand that they do the identical.
Stephen Embry is a lawyer, speaker, blogger, and author. He publishes TechLaw Crossroads, a weblog dedicated to the examination of the strain between expertise, the legislation, and the observe of legislation.
The publish In-House And AI: You Can’t Demand What You Don’t Understand appeared first on Above the Law.