AI for Lawyers: Why Your Judgment Matters More Than Ever (featuring Colin Levy)

Colin Levy has read more legal AI hype than almost anyone, and he thinks most of it is useless. In this episode, Jonathan Cullen sits down with the General Counsel of ‪@hellomalbek‬ and author of the new book Code Switched: A Lawyer's Guide to the Era of AI to cut through the noise around AI for lawyers. Colin explains why he thinks lawyers should stop saying please to ChatGPT, why AI hallucinations are a feature not a bug, and why the real risk in legal AI adoption isn't the technology replacing your judgment, it's you forgetting to use it. They get into tokens, AI agents, the billable hour, and why Colin believes we're only in "chapter one" of what artificial intelligence will do to the legal profession. If you've been putting off learning how AI works in legal practice because you're not sure where to start, this episode is your starting point. Guest Bio Colin Levy is General Counsel at Malbek, and one of the most respected voices in legal technology and AI for lawyers, with a large following on LinkedIn and through his newsletter. He's an adjunct professor at Albany Law School, author of The Legal Tech Ecosystem, and editor of the Handbook of Legal Tech. His new book, Code Switched: A Lawyer's Guide to the Era of AI, is on presale now and releases October 6, 2026. Timestamps 0:00 Intro: AI for Lawyers with Colin Levy 1:00 Code Switched: What the Book Title Means 3:00 Writing an AI Book That Stays Relevant 6:00 Global AI Regulation for Legal Teams 8:00 Why Bjarne Tellmann Wrote the Foreword 9:00 Best Chapter for AI Skeptics 11:00 ChatGPT Mistake: A Real Legal Example 12:00 How to Prompt AI Like a Lawyer 13:00 Why AI Hallucinates: Tokens Explained 15:00 AI as a Thought Partner, Not a Chatbot 16:00 AI for Lawyers: Judgment Over Replacement 19:00 The Chainsaw Analogy: AI and Legal Judgment 20:00 Stop Saying Please to Your AI Tool 23:00 Why AI Tools Are People Pleasers 25:00 AI and the Billable Hour Model 27:00 Legal AI Adoption: We're in Chapter One 28:00 AI for In-House Legal Teams 29:00 Week One of Responsible AI Adoption 31:00 Why AI Rollouts Fail Without Training 34:00 AI Hallucinations: Feature, Not a Bug 37:00 Lateral Thinking and AI Creativity 39:00 The #1 Mistake Lawyers Make With AI 41:00 AI for Solo and Legal Aid Lawyers 42:00 AI Ethics for Lawyers: The Real Question 44:00 AI Fails in Law: Real Examples 46:00 Disclosing AI Use to Legal Clients 47:00 AI Agents Explained for Lawyers 49:00 Fruit of the Poisonous Tree: AI Agents 51:00 Legal AI in 5 Years: Read the Book or Not 52:00 What Writing Code Switched Taught Colin 53:00 Rapid Fire: AI Advice for Lawyers 55:00 Where to Find Colin Levy and Code Switched Links Colin's website: colinslevy.com Colin on LinkedIn: linkedin.com/in/colinslevy Colin on Twitter/X: x.com/Clevy_Law Colin on Instagram: instagram.com/clevy_law Code Switched: A Lawyer's Guide to the Era of AI (releases October 6, on presale now): On Amazon & here: https://amplifypublishinggroup.com/pr... If this episode stuck with you, take the Free Lawyer to Leader Assessment at jonathancullencoaching.com Key Takeaways Code Switched is about the shift from AI helping lawyers finish work to AI helping lawyers build things, a bigger adaptation than any previous legal tech shift. AI for lawyers means AI amplifies judgment, it doesn't replace it. The more powerful the tool, the more skill you need to wield it. Colin's line: a hammer makes it easier to build a wall, but you still need to know where the nail goes. AI hallucinations are a feature, not a bug. Large language models are predictive, token-based systems. They match patterns, they don't understand your question. That means wrong answers aren't a glitch to patch, they're baked into how the technology works. Stop humanizing AI tools. Don't say please, don't treat "I" and "me" as anything but a marketing choice. AI models are built to be people pleasers, which means the lawyer has to be the skeptic in the room, every time. Responsible AI adoption in legal practice starts small and low-risk. Week one should mean one repetitive, low-stakes task done in a controlled way, not a firm-wide rollout with no training behind it. The real AI ethics question for lawyers isn't "should I use AI." It's "how am I using it." Client confidentiality, data hygiene, and transparency matter more than the yes-or-no debate most people are stuck on. Shareable Moments "Assume it knows nothing, and assume you need to give it a seemingly absurd level of context." "These tools don't understand the way humans understand." "AI doesn't power anything." "We're kind of in chapter one. We finished the prologue." "Hallucinations... are a feature, not a bug." "You shouldn't be worrying about hurting its feelings. It has no feelings." "That old saying, it seems too good to be true, can often be the case" "The number one mistake most lawyers make when using one of these tools is assuming it's just gonna be right"