Law firms must proactively redesign new attorney onboarding to intentionally build both AI fluency and power skills now, rather than waiting for AI's disruption of traditional training grounds to settle into a steady state.
Listen to this article
Key insights:
- Associates form AI habits regardless of firm guidance — Early-career attorneys are already using AI tools independently, whether firms train them or not, creating urgency for firms to shape responsible use before informal habits become entrenched.
- Structured, iterative onboarding replaces one-time training — The model being used by law firm Duane Morris, which includes combining extended orientation sessions, designated "AI champions" at the practice-group level, and traditional mentorship, shows that effective onboarding requires multiple overlapping pathways.
- Power skills, not technical AI skills, differentiate top talent — Because AI has lowered the bar for surface-level competence, human capabilities like instilling client confidence and business fluency now matter more in distinguishing "good" from "outstanding" lawyers.
It is common knowledge to talent leaders within law firms that the training ground that built legal judgment is disappearing faster than firms can adjust to it. AI is now absorbing the tasks that built associates’ capability as competent lawyers; and yet, entry-level attorneys joining new firms are still counting on their employers to develop them into trusted lawyers.
Diane C. Downs, Chief Legal Talent and Growth Officer at law firm Duane Morris, is not waiting around to see how this all plays out. "I do think we've seen these [technology] transitions before, and we've had these same worries," Downs says. "But the difference with AI is that it's happening much faster and with a lot more iteration."
Indeed, the speed of change is why onboarding of entry-level associates has to be rebuilt now during mid-transition rather than waiting for things to settle into a new stable pattern.
Associates are already forming AI habits
A complicating factor impacting responsible use of AI is that many future and new lawyers have been using AI tools for years and will do so even in the absence of rules. This is creating urgency for many law firms in regard to revamping onboarding programs.
For example, Downs recalls that when attorneys first tested ChatGPT, mid-level associates warned each other to avoid using it because they were convinced it would create more work for them when they reviewed junior associates’ work. Four years later, the same associates are building their own agents and automating repeatable tasks with or without their firm training them to do it well.
And while there is great training for attorneys on what constitutes responsible use of AI, such training must include verification steps and human review to be truly effective.
While the firm leaves it up to the lawyers to decide how to use AI in the creation of first drafts, the firm has one fixed principle: A first draft is never an appropriate final submission.
Law schools saw the same compression. There was outright resistance to AI use among law schools up until 2023, Downs notes, and now, that has given way to AI being built into curricula by 2024. In fact, the resistance created a cohort of students and young lawyers who often knew the tools better than their professors and seasoned colleagues.
This pattern reinforces the argument for redesigning onboarding now. Early career attorneys do not wait for an institution’s official position before forming habits with an AI tool. Waiting causes a risk for firms that by the time leadership settles into its own comfort level with AI, many associates have often already built their working habits with it.
What a redesigned onboarding program looks like
Structurally, Duane Morris has made several changes to its onboarding programs. New lawyers get a two-day, in-person orientation in Philadelphia, including a live AI-and-library research session. After the initial orientation, Duane Morris inserted twice monthly extended orientation sessions that run from September through the following June and feature key internal staff and outside experts as speakers.
These sessions sit alongside existing training on key legal skills, which include effectively legal writing, workplace navigation, and early marketing, and networking guidance. Each summer, her team debriefs the outgoing class to redesign the program for the next one.

The process of learning legal judgment and effective AI use occurs in their first year through multiple pathways. While the firm leaves it up to the lawyers to decide how to use AI in the creation of first drafts, the firm has one fixed principle: A first draft is never an appropriate final submission, according to Downs.
At the practice group level, each practice has designated AI champions, which is usually a partner, counsel, or senior associate who has early access to test tools against real work. This way, the champions, can figure out the most effective way to utilize each tool. These champions then help new lawyers and others to use practice-relevant AI tools effectively and efficiently to avoid every lawyer having to learn how to use them through individual trial and error.
A separate, traditional mentorship track remains in place to give new associates a resource for navigating cultural questions, including AI-specific ones, such as whether using AI with a particular partner will be read as inefficient.
Raising the bar for power skills
Part of the restructuring of onboarding also means increasing the firms' focus on cultivating so-called power skills, which are often called soft skills — however, using the term power skills reinforces their importance in the AI age.
Indeed, these capabilities are more important currently because it is easy to demonstrate surface-level competence with AI use, and power skills are increasingly a differentiating factor in determing whether someone is a good lawyer or an outstanding one.
In particular, Downs cites two areas of power skills that are critical for lawyers: i) the ability to instill client confidence, especially when delivering bad news; and ii) holding a depth of business fluency that turns a lawyer into a trusted advisor. Downs' sharper point, however, is that AI has made it harder for those outstanding lawyers to stand out. “Even the most awkward lawyer" can now ask a tool to draft talking points, which raises the baseline everyone is judged against, she adds.
While there is great training for attorneys on what constitutes responsible use of AI, such training must include verification steps and human review to be truly effective.
Importantly, the use of AI simulations has emerged as a key opportunity to instill power skills earlier in attorney development as part of firms’ onboarding program. At her prior firm, for example, Downs says she tried to build these skills deliberately using simulations, but she saw mixed results. More specifically, she helped test an avatar-based tool for practicing difficult feedback conversations in which a human coach controlled the avatar in real time. Very few people utilized the tool.
Years later, a Chat GPT-based tool for the same use case was rebuilt under her leadership, and the tool succeeded. People unwilling to rehearse with a live coach engaged with the AI version privately. Downs says she is now in early talks with a former colleague, who built a similar bedside-manner simulator for physicians, about adapting the concept for attorneys.
If simulation can show how power skills can be developed on purpose instead of leaving it to chance, then we should ask whether firms will make that same intentional choice regarding how to use AI effectively in their onboarding programs as a whole.
Nothing about AI holds still long enough for a firm to get to a steady future state with it. And for many law firms, the approach used by Duane Morris to iterate attorney development initiatives, including onboarding of new lawyers, may be the most meaningful way to move forward.
"We are engaging and having to figure out the new way to practice law," Downs says. "And those that engage with [AI] and continue learning are the ones that are going to be successful using it."
In an AI age that refuses to sit still, the willingness to keep learning is the one skill for which no tool can substitute.

