Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Pricing

Joel Hron

👤 Person
406 total appearances

Appearances Over Time

Podcast Appearances

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

But they're sort of inviting you in to make the hard judgments at the right moments.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And those are like the most human and the most challenging parts of the job.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

So I think what you'll see is that really the best attorneys are the ones who know how to use these tools.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

The best will actually become a hundred X of where they were versus, you know, just kind of leveling up everybody at the same time, if that makes sense.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

Yeah, it's a good question.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And I think it's a pattern that's still evolving, quite honestly.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

I think Andrej Karpathy gave this talk at Y Combinator a few weeks back where he talked about the generation verification loop of AI in humans and really designing user experiences that spin that flywheel of generation and verification as fast as you can.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And that's really how you get tremendous speed up in value at the end of the day.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And so I think, frankly, some of these design patterns exist in our products.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

Some of them are still being evolved at the moment.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

What I will say, like the current UX of research, for instance,

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

you know, shows the user like the trajectory of steps that the agent is taking, what searches it's doing, what it learned, why it's doing this next search or why it's going to this content repository versus this one.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And it sort of spells that out and it does that in real time, but it also does that at the end.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And you can look at like, these are the notes that the agent has written for itself over the course of doing this research.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And here's how it got to this answer.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

What I would love to see in the future is users being able to inject themselves either after the fact or even in real time to say, oh, no, I've seen that case before.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

I think you should check this one out instead.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And I think these little redirect nudges and things like that add context to the agent as it's going through its process and are good ways for a really 10x lawyer, for instance, to

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

to really, you know, apply what they know and their experience and maybe their broader context of the case and the client and the judge and all this stuff to help the model get to the right answer or the most right answer that it possibly can.

The Neuron: AI Explained
How Thomson Reuters Built AI Agents That Think Like Lawyers

And so I think, again, some of these patterns exist already in the UIs, but I think, you know, we're certainly eager to, like, continue to evolve more of them as we go forward.