The Innovation Bottleneck No One Talks About
With new associates and summer programs underway, many firms are asking a timely question: Should associates be using generative AI?
The answer isn’t straightforward—because the stakes are high. Technology is evolving fast. So are expectations around associate development, billable work, and client value. Legal professionals are caught in the middle: they want to stay relevant but may be unsure how to use tools that still feel experimental.
So perhaps the better question is: What kind of use, under what conditions, actually helps associates contribute more and learn faster?
A Real Question for a Real Shift in Legal Practice
Firms are investing heavily in tools like Harvey, CoCounsel, Lexis+AI, and Copilot. But adoption remains uneven. New attorneys often arrive without clear guidance. Partners are still deciding what "appropriate use" looks like. And leadership is under pressure to show ROI without sacrificing quality or security.
That’s why many are starting here: should associates use generative AI tools to support their daily work?
The Learning Dilemma
A common concern is that if associates rely on AI, they won’t truly learn the craft. Deep thinking, the argument goes, comes from grinding through research and wrestling with messy first drafts.
That fear is understandable. Giving an associate GenAI can feel like handing them a race car right after passing their driving test. Without experience, how can they spot hallucinations, oversimplifications, or flawed reasoning? And if partners don’t understand how the output was created, how can they coach improvements?
But here’s the reality: GenAI already produces research-driven outputs that match - or exceed - the quality of first- or second-year associate work. That’s not hypothetical. It’s happening now.
So, where does that leave us?
For Partners: Don’t Delegate What You Haven’t Explored
It's tempting to say, "Let the associates figure it out." But that rarely works.
Partners have a responsibility to understand the basics of GenAI tools. With short-form programs like GenAI Edge, there's no excuse not to build at least foundational familiarity.
Why? Because delegating without understanding leads to:
- Mismatched expectations
- Shallow adoption
- Missed learning opportunities
Partners who engage alongside associates can:
- Provide practical guidance
- Spot errors and potential risks
- Support associates in developing legal judgment
Yes, there’s a fair concern: if GenAI drafts the work, what happens to deep training?
The solution isn’t to avoid AI. It’s to evolve how we teach and supervise. According to Wolters Kluwer (2024), 67% of partners worry that GenAI could erode traditional learning paths.
The Case for Associate Use: Learn Faster, Think Deeper
Used intentionally, GenAI helps associates:
- Draft more quickly
- Explore diverse perspectives
- Iterate toward a stronger work product
There are many tools in play - ChatGPT, Claude, Copilot, and legal-specific platforms like Harvey, CoCounsel, and Lexis+AI. Today, most apps have some form of GenAI built in.
What matters isn’t which tool you use, but how you use it:
- Know what tools are available
- Match the tool to the task
- Ask better questions
- Evaluate outputs critically
- Synthesize, not just accept
- 30% productivity gains
- 25% less time on routine drafting
- 20% higher accuracy in document review when AI was used with clear task framing
For associates, the future of legal practice will involve daily use of GenAI. The real question is how to ensure learning keeps pace.
Learning Through AI: What Engineering Taught Me
My background in chemical engineering offers a helpful analogy. When computer simulations first emerged, veteran engineers worried the next generation would lose touch with the fundamentals. They feared we’d stop learning complex math, since software could now run calculations instantly.
But something different happened. Instead of calculating one perfect answer by hand, we ran dozens of simulations. We explored edge cases, stress-tested designs, and gained insight through comparison and iteration. That’s how intuition and expertise were built.
Did our math skills decline? Yes. Did the quality of work suffer? Not at all—because different, equally valuable skills emerged.
Law can evolve the same way—if we’re intentional.
The 5× Memo Method: Better Thinking, Faster
Here’s a framework for using GenAI as a learning engine, not a shortcut:
- Identify 3–10 distinct angles for the memo (e.g., legal, client risk, reputational, precedent, tone)
- Select 1–3 GenAI tools to work with
- Generate 2–5 drafts using GenAI, one per angle and tool
- Prompt the tool to evaluate the strengths and weaknesses of each
- Ask for a merged or improved draft that addresses the identified gaps
- Manually refine the output into a final deliverable
- Reflect on what you learned
This may sound like a lot of work, but it often takes less time than drafting a single memo from scratch, and it supercharges learning:
- Highlights multiple lines of reasoning
- Shows how framing affects content
- Builds critical thinking through structured comparison
As the Stanford Humanity Center puts it:
"We need to change this mindset to cultivate a workforce that can successfully collaborate with intelligent machines. The risk is a world without work; the reward is a vibrant and curious new humanity."
What’s at Stake
McKinsey (2023) estimates that 23% of legal work is automatable today - much of it the work associates currently do. That’s a threat if we ignore it. But it’s also a powerful opportunity.
Letting associates use GenAI without support? Risky.
Refusing to let them use it at all? Even riskier.
The best path forward is structured experimentation, collaborative learning, and rethinking how legal judgment is built.
Because GenAI won’t replace associates, but it will redefine what makes them exceptional.