ILTA’s EVOLVE Conference got off to a rousing start with an opening keynote address by Zach Abramowitz. Abramowitz is a real thought leader in the legal tech and innovation field. He is also a legal tech investor, consultant, and GenAI evangelist. Provocative. Challenging. Rapid fire. And different.

The title of his talk was Most Law Firms Are Doing AI Wrong. Here’s How to Do It Right. The basic premise was something I have observed: too many law firms are rushing to buy GenAI products (think FOMO) without understanding them. As a result, they can’t employ them for the highest uses.

And while I agree with Abramowitz that the power and potential change from GenAI is perhaps unprecedented, the failure of law firms and lawyers to understand new technologies and what they can do isn’t.

When law firms first began mandating that computers be placed in lawyers’ offices, I remember walking into a partner’s office and observing his computer was off. When I asked Sam if he had been using it, his response was the damned thing never comes on.

A Fundamental Disconnect

It’s much the same today: there’s this rush by law firms to grab GenAI and compel every lawyer to somehow use it without grounding them in what the technology is and how it works. Abramowitz believes this results in a whole host of questions and problems.

Questions and problems such as training that does not seem to be effective, failing to understand how to best use and assess the real ROI, concerns whether GenAI will end the billable hour and the role of lawyers, what to tell clients, and what new and expensive tool to buy, among others.

It is a proverbial fundamental disconnect.

It All Starts with Thinking Differently

Abramowitz believes all these issues can be addressed by starting from an understanding of how GenAI works. And removing a misunderstanding that is either overly simple or overly complex.

The overly simple misunderstanding is based on the idea that GenAI works by just predicting the next word based on data. In fact, it’s not just predicting the next word but understanding the context in which a word is used. Based on that understanding of context and the relationship between all the words and contexts, it comes up with a response. My phone can predict the next word I might want to use in a text but it can’t answer an inquiry like Claude or ChatGPT does. That’s the oversimplification side.

The overly technical misunderstanding is based on treating GenAI like other technologies and the idea you have to jump through certain hoops to make it work. It’s the notion that assumes you can’t get a result unless you go through some complicated steps applying a mysterious concept of prompt engineering.

But Abramowitz says the difference in GenAI is that it understands language much in the way humans do. I recently wrote about a friend who told me all you have to know to use GenAI is that there is not all that much to know. You just talk to it, use it, and work with it. It’s a paradigm shift, says Abramowitz. It’s a different way of thinking.

Abramowitz says training should focus not on oversimplifying or overcomplicating GenAI but in understanding what it does and what it can do and that it is rapidly evolving. This is where law firms need to focus.

The Hallucination “Problem”

The big GenAI concern is hallucinations. But if you understand how it works, you know hallucinations are not a glitch, but a feature, as I recently talked about. It’s built in to how it works. It uses imperfect information to try to connect dots, much as we humans do. And just like us, it can make mistakes. Abramowitz gave this example: if you tell a GenAI tool to say I don’t know when it isn’t sure, you risk getting the answer I don’t know to every question.

Knowing this, and as I have written, knowing when a GenAI tool may be likely to hallucinate may make all the difference.

Real Use Cases

Which leads to another key Abramowitz point: knowing how GenAI works can lead you to focus on the best use cases. In fact, Abramowitz believes the concept of best use case is wrongheaded. Instead, the focus should be on using it for the highest performance cases, he says.

What are these? Abramowitz argues it’s things like using it as a second brain. Using it to learn what you don’t know. Using it to pitch business, prepare presentations. Even using it to understand how AI works and what it can do.

But there’s more. Abramowitz contests the prevailing notion that AI should be used for mundane work so that humans can do the high-end, strategic work. That stems from a misunderstanding of what GenAI can do. What it is really good at are things like brainstorming. Strategic thinking.

And, says Abramowitz, it’s decidedly not good for just doing your everyday job. Instead, it should be used to do things that you didn’t think possible.

That leads to the understanding that, based on what the tools can do, the ROI is not necessarily increased efficiency. It’s the improved result, increased capabilities, improved branding, and ability to price competitively.

Legal Research?

I was talking to someone about this notion after the talk within the context of legal research. It’s in the legal research realm where we see the very public hallucination problems. But if you follow Abramowitz’s logic, perhaps legal research falls within the realm of the more mundane kind of work. The high-end work, where GenAI can really help, is in applying what the researcher finds to the problem at hand. It’s asking, given the law, what are the ways the clients’ desires can be accomplished.

But somewhat paradoxically, that may require more work and more critical thinking on the front end to get to the best uses of AI on the back. I fear that too many will use it to short cut this hard work.

(I can feel the shivers of several vendors down my spine.)

Dealing with Younger Lawyers

It also means adjusting the view of the need for younger lawyers. If we understand what the tools can and can’t do and train younger lawyers on this, a younger lawyer can gain the expertise of a more senior lawyer in a fraction of the time. I also agree with Abramowitz: if I had had GenAI tools when I was a younger lawyer, I would have avoided a lot of sleepless nights and produced better work.

The key, says Abramowitz, once you train folks, is to understand the tool, is to encourage them (and yourself) to use it. It’s not about picking the perfect tool; it’s about avoiding analysis paralysis. I get a lot of use out of GenAI not because I took any prompt engineering training but by understanding it and rolling up my sleeves and using it.

A Critical Question

Which leads to another Abramowitz core concept: we need to stop asking only what GenAI can do. We need to ask, says Abramowitz, because we have it, what us humans ought to be doing. It’s asking why we need certain things or are doing certain things when GenAI can supply or do those things better.

It’s asking what will happen because of GenAI. Questions like what the role of a law firm is when much of what it now provides can be done with AI tools and lawyers can practice more or less on their own.

Some Real Provocative Thinking

Abramowitz offered some provocative points on the future. He rejects the notion that GenAI will bring about the end of the billable hour. Instead, he thinks those lawyers who use GenAI effectively will be able to raise their rates significantly because they get the best results in a competitive marketplace.

And, as he has talked about before, it won’t reduce the amount of legal work. It will greatly enhance it. New issues, new concepts, new problems. As he put it, do you think the development and building of all these data centers is not going to increase legal work?

Abramowitz says GenAI is not going to replace lawyers but is going to raise the standard of what is good work. Those who use it to do better work, will get more work, just like always.

But It All Stems from Something Basic

All of this and much more were covered in Abramowitz’s one-hour presentation. But it all comes back to one point: the need to focus on how GenAI works and neither oversimplifying nor making it too complicated. It’s like Sam and his computer that wouldn’t turn on. The problem was that no one talked to him about how computers work. What they can do and what they can’t. What are the highest performing uses for them.

And understanding that for it to work, you have to start by turning it on and using it.

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.

The post The Provocative Abramowitz Keynote And The Computer That Won’t Come On appeared first on Above the Law.