ChamberOne

Change management when partners resist AI

By the ChamberOne team 4 minute read

In many firms the loudest voice on AI is not the enthusiast but the partner who will not touch it. Managing partners often treat that partner as an obstacle. A better view is that resistance carries information. Some of it points to real risks that the enthusiasts have not thought through. Some of it reflects fears about fees or status that deserve a straight answer. This article is about moving a firm forward without a fight, by listening first and setting up the change so that sceptics have a part in it.

Key points

  • Listen to each sceptic first; their objections usually point to a specific outcome they fear.
  • Sort objections into risk, fees and identity, and answer each differently.
  • Pilot with willing teams, record results honestly and report in terms partners care about.
  • Ask sceptics to help write the guard-rails, and give every pilot an off switch.

Listen before you plan

It is tempting to arrive with a deck and a vendor. That tends to harden positions. Start instead with a conversation with each sceptical partner. Ask what worries them, what they have heard, and what would have to be true for them to be comfortable. Write the answers down and use their words when you report back.

You will usually find that people are not against technology as such. They are against a specific outcome: a client complaint, an embarrassing mistake, a fall in fees, a loss of the craft they spent years learning.

Three kinds of objection

Most resistance falls into three groups, and each needs a different response.

  • Risk: confidentiality, accuracy, regulatory exposure and professional liability. These are legitimate. Answer them with controls, policy and honest limits, not reassurance.
  • Fees: the fear that faster work means lower income, or that clients will demand discounts. Answer with an open discussion of pricing, scope and what the firm will do differently.
  • Identity: the sense that drafting and research are what a lawyer is, and that a machine doing them diminishes the person. Answer with respect, and with a picture of how judgment and client relationships become more visible, not less.

Take the risk objections seriously

Sceptical partners are often right that unchecked AI output is dangerous. Fabricated authorities, misquoted clauses and leaked data are real problems. Do not wave them away. Agree that these risks exist, and show what the firm will do about them: approved tools, a list of what may never be entered, verification rules and a log.

The strongest move is to ask the sceptic to help write the rules. A partner who has pushed for strict verification will defend that rule in front of colleagues more convincingly than any manager.

Start with the willing

Do not try to convert the whole firm at once. Find a team or two with a clear, repetitive task and some curiosity: a litigation support group sorting documents, a conveyancing team reviewing leases, a finance team checking invoices. Give them a small, defined pilot with a start and end date, and a few things to measure.

Keep the pilot honest. Record the time taken before and after, the corrections made by reviewers and the problems encountered. Share the bad news with the good. A pilot that hides its failures will lose credibility the first time a sceptic finds one.

Choose a pilot lead who is respected by colleagues, not necessarily the most enthusiastic person. A careful senior associate who reports plainly on what went wrong is worth more than a champion whose reports are always glowing.

Show results in the partners' terms

Partners care about client satisfaction, quality, profit and risk. Report in those terms. A story such as 'the team finished the review in two days instead of four, the associate caught three errors in the tool's draft, and the client was told how the work was done' persuades better than a general claim about productivity.

Avoid made-up numbers. Use your own firm's records, even if small, and say how they were measured. If the results are mixed, say that. Sceptics trust people who admit limits.

Make it safe to try and safe to stop

Resistance eases when people know they can step back. Give every pilot an off switch and a named person who can use it. Agree in advance what would end the trial, such as a confidentiality incident or repeated serious errors. Offer training that is practical and short, with real examples from the firm's work, and let people practise on non-client material first.

Be careful about pressure. Telling partners they will be left behind breeds resentment. Offer help, time and a way of trying the tools without public embarrassment.

Look at incentives and pricing honestly

If the firm bills mainly by the hour, then faster work can feel like a pay cut, and resistance may be rational. Say so, and discuss how pricing might change: fixed fees for defined tasks, scoped packages, or different treatment of review time. Check how clients are likely to respond and how your regulator or bar treats fee arrangements.

Consider how you reward the people who make the change work. If associates who find errors in AI output are seen as slow, the checks will fade. Recognise careful verification as part of quality.

Keep going without declaring victory

Review the policy every few months. Invite the sceptics to the review. Publish what the firm has learned, including the mistakes. Some partners will never use the tools, and that may be acceptable, as long as they do not block others and they are not left unaware of what the firm is doing. Success is a firm that uses AI where it helps, checks it where it matters and can explain both to a client.

Share a short note each quarter with all partners: what the firm tried, what it learned, what it stopped and why. People who are not using the tools still see that the firm is handling them with care, and the sceptics' early warnings are visibly taken into account.

  1. 1Run a pilot with a willing team
  2. 2Record time, corrections and problems
  3. 3Report results in the partners' terms
  4. 4Review the policy, with sceptics invited
  5. Then back to step 1.
A pilot and review loop

This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.

Read next

Change management when partners resist AI | ChamberOne