How legal professionals are using AI with Benjamin Softness, partner, King & Spalding LLP

Artificial intelligence is rapidly transforming the practice of law, from legal research and document review to case strategy and client communications. But how are California's legal professionals actually using these tools in their daily work? In this series, the Daily Journal speaks with judges, attorneys, mediators, support staff, academics and other legal practitioners about their experiences with AI -- exploring what's working, what concerns them, and how they're navigating the ethical and practical challenges of integrating new technology into a profession known for its adherence to tradition.

What specific task or workflow in your practice, chambers, or studies has AI changed most dramatically, and what does that look like day-to-day?

The workflows that have changed the most for me are at the top of the funnel, the beginning of a long process. For example, when I first start on a matter, I often ask AI to review the case-initiating document - for instance a complaint or a subpoena - and summarize the document and extract key issues or nuggets of information. It's not a substitute for my own review, but it provides instant context and informs my own review.

When you're using AI tools for legal research or drafting, how do you verify the output? What's your process for catching hallucinations or errors?

I do not use AI for legal research. One of the most critical skills lawyers must bring their clients in the context of AI is an understanding of what it is good at and what it is less good at, at least today. Lawyers are far better than AI today at finding authority and sensitively assessing its persuasiveness. I focus my own use of AI on what I call "closed-ended" (as opposed to open-ended) tasks, such as summarizing, extracting, sorting, or comparing - all with respect to authorities or documents that I provide the tool.

Have you encountered a situation where AI led you astray or gave you problematic advice? What happened and what did you learn?

AI can sometimes overread a document or confuse issues. The lesson, always, is to check the work. Some tools make that easier by providing citations for the conclusion the tool is reaching. Reviewing these citations is always a must. Consider pressure-testing AI by asking it to try to disprove what it just told you.

How are you thinking about confidentiality and data security when using AI tools? What guardrails have you put in place?

Confidentiality and security are non-negotiables for me. Our firm has a robust governance program for the use of AI, the central pillar of which is limiting AI use to tools the firm has vetted and approved for confidentiality, data security, and other requirements.

What kind of legal work do you think AI will never be able to do well, and why?

Never say never, but has AI ever taken a client out to lunch? Jokes aside, there many things AI can't do well, and shouldn't. Our judgment is our stock in trade. AI can help assess risk or compare the puts and takes. But lawyers can advise. Advising takes judgment - awareness of a range of substantive and fundamentally human factors, plus a willingness to take a position on a close call. In addition, lawyers today are far better than AI today at finding authority and sensitively assessing its persuasiveness along a wide range of human dimensions. For instance, how persuasive will this authority be in this case with this judge? I think these measures of judgment are here to stay, at least for a while.

How has AI affected your professional in terms of what services or work you provide, how you communicate, or what others expect from you?

I think of AI as adding a capability. Now, clients and colleagues know I have another club in my bag. In particular, sometimes clients need speed over perfection (but, of course, sometimes they need both!). When speed is the ask, starting with AI can be a massive and previously unavailable head start.

If someone just entering the legal profession asked you how to think about AI in their career development, what would you tell them?

It is no substitute for learning, writing, and understanding. It can be tempting to rely on AI for everything. Resist that temptation. Knowledge and experience inform judgment. And remember: if you let AI do the things you're supposed to be good at, you're engineering yourself out of demand.