CBA Record July-August 2026

good advice for any technology, but espe cially with AI. It’s important to find one thing you can improve that’s small and start there, as opposed to tackling a huge project out of the gate. Get some small process improved, increase your comfort level, and then work upwards from there. Other sessions addressed specific implementation questions such as draft ing with AI, litigation risk, e-discovery, due diligence, negotiation support tools, use in M&A, use in investigations, jury consulting, and case strategy, and sev eral vendors had tools that support these areas. In the contract drafting session, for example, one tip was to start with high volume, repeatable contracts. Because it can be a bit risky to trust AI tools with bespoke, novel, or high-consequence pro visions, another approach is to start with simply using these tools for issue spotting, then move up to having the AI suggest certain provisions to be made, before moving all the way up to having AI gener ate a first draft. As for preventing halluci nations, when conducting quality review, make sure that the AI is citing a proper case for the proposition, not just that the case name and citation are correct. The intellectual property panel had different concerns with AI as it focused on a specific area of law. Panelists addressed broader IP concerns, including whether AI-generated outputs are protectable under copyright and patent law. (Typical lawyer answer—it depends.) Case law is clear under copyright and patent law in that a human needs to be involved in the creative process, but the amount of that involvement is still a fact-specific question. Overall, the symposium was a great event. It cemented the CBA as taking a leadership role in the discussion around AI, especially with the focus on practical, ethical implementation. Last word: Every practitioner using AI tools needs to keep the Illinois Supreme Court’s Policy top of mind.

LPMT BITS & BYTES BY KEVIN THOMPSON AI 2035 Takeaways: Optimizing AI for Your Legal Practice

T he CBA’s AI 2035 Symposium was packed with great takeaways for lawyers, judges, and law stu dents. Many of the sessions provided thoughtful guidance framed by the Illi nois Attorney’s Guide to Implementing AI, last year’s ARDC release supporting the Illinois Supreme Court’s Policy on Artifi cial Intelligence. Many vendors presented helpful tools, some of which I use in my own practice. The “Vendor Selection and Risk Assessment for AI Tools” panel was helpful for looking more carefully at the ARDC guide. Many tasks can be handed off to AI for assistance, rang ing from initial document drafting, case intake and assessment, contract analysis, legal research and cite checking, sum marization, document classification and analysis, chronology generation, privilege flagging, to timekeeping. Dozens of vendors are out there to assist with these tasks; some are legal specific and others are not. Your duty of technological competence doesn’t mean you have to understand all the details, but you do need to have a general under standing so you know what AI is doing. A good analogy is that you don’t need to be able to reprogram a car’s onboard computer, but you need to know what the dashboard lights and warnings mean so you know when the car needs attention. Some good questions to ask a potential

vendor include: Does your vendor have legal industry experience? Is the product more than a pretty wrapper put on Chat GPT? What support does the vendor offer? What happens when something goes wrong over the weekend and the brief is due on Monday? What training data was used? How does the vendor account for hallucinations? The symposium’s administrative focused track featured two comple mentary sessions. One discussed the “Collapse of the Billable Hour” caused by AI implementation. The other panel discussed budgeting and the return on investment for AI implementations. The analysis for one necessarily leads to the analysis for the other. Just because it used to take 10 hours to do something without AI help doesn’t mean you can still bill the client for 10 hours (that would be unethi cal if it didn’t take that long). However, you could implement a flat rate for that service that is priced appropriately. You need to know the real costs of the plat form to make that calculation properly. Both panels stressed the importance of knowing real numbers. Some of the best sessions provided practical tips for implementing AI. One vendor, Framework IT, outlined what they call the “Crawl, Walk, Run” plan for implementation: Start slowly with simple tasks, then work your way up to implementing a larger solution. That’s

Kevin A. Thompson, a self-described geeky lawyer and former chair of the CBA’s Law Practice Management and Technology Committee, heads Levin Ginsburg’s intellectual property and data privacy practice. He receives no compensation from vendors mentioned in this column.

48 July/August 2026