November 18, 2026

Corporate legal departments increasingly expect their outside law firms to use artificial intelligence and lower their fees. Yet many in-house teams are still struggling to evaluate the technology themselves, raising questions about how effectively they can hold firms accountable.
A survey sponsored by alternative legal services provider Axiom found that 95% of respondents expect outside counsel to use AI, while 92% expect lower rates when firms do. A separate survey cited by Above the Law found that 80% of in-house counsel neither require nor encourage their firms to use the technology.
The Axiom study suggests that a lack of practical experience may help explain the gap between expectations and demands. InsightDynamo conducted the online survey of 528 in-house legal professionals at companies with annual revenue of at least $100 million between March 13 and March 25, 2026.
Axiom supplies legal talent to corporate departments and has a commercial interest in businesses seeking outside help with AI. That interest does not invalidate the findings, but it is relevant context when considering the results.
Among respondents, 83% said they did not know how to measure AI’s return on investment or assess the effectiveness of their department’s spending. Only 7% reported optimizing and measuring AI across their organization.
Skills and regular use also appeared limited. Just 10% considered their AI skills advanced, and 25% said their departments sent personnel for external training. Only 17% reported that a majority of their workers used AI tools daily. Meanwhile, 40% had implemented AI planning and governance practices, and another 40% said they planned to do so.
Respondents identified several obstacles to determining AI’s value, including concerns about disruptive workflow changes, employee resistance, insufficient skills and the difficulty of measuring returns.
The survey also found shortcomings in how departments select tools. Only one in 10 teams conducted a rigorous evaluation before buying a product, and fewer than one in four ran structured pilots with defined goals, timelines or measurements. Two-thirds used general-purpose or frontier models, such as ChatGPT, Copilot and Gemini, rather than legal-specific tools.
Nearly all respondents—98%—said they wanted outside guidance on selecting tools. A further 96% said they would use an alternative legal services provider for adoption or implementation advice rather than a law firm.
The findings point to a challenge for legal departments seeking savings from outside counsel. Firms that bill by the hour may have little financial incentive to reduce the time spent on work, while clients without clear measures of AI’s benefits may struggle to specify where the technology should be used or how savings should be shared.
For in-house teams, developing those measures may be a necessary first step toward turning broad expectations into concrete demands.