Mike Mace asked Orlova what level of trust people should have in an AI system and how to measure it. She describes a test with over 200 participants where one assistant confidently answered an impossible pricing question in two-thirds of cases, with answers ranging from $1 to $8,000, a failure no single user could see from inside the session.
…experiences. And so that's why we needed to come up with a new framework, new approach, still pulling from some of the incredibly important aspects of it, quantitative and qualitative, a consistent measurement tool, but have something where we can test AI experiences in a same and similar fashion.
…test was the number of times people came back to the idea of trust. Do people trust AI? How do we measure trust, et cetera, et cetera, et cetera? Can you talk about what's the appropriate level of trust and the type of trust that people ought to have in an AI system? And how do we measure that?
Yeah, that's trust is difficult, right? Almost everyone frames trust as something that they want to maximize. More trust, better product, better experience. I don't agree with that when it comes to AI. What you want is an appropriate level of trust. You want an accurate trust meter. People rely on AI when it deserves it, but then they step in when it doesn't do something right. Too little trust and the product doesn't get adopted. People just use it once, they don't believe it, and they go back to doing the things that they used to in the past. Too much trust can be very, very dangerous. And not many people talk about it. Not many people measure it. If someone accepts an answer that is actually wrong, and then if there are consequential decisions that they make based…
I want to be mindful of time because we've got a certain amount of time we can use for the podcast. But Sarah, I'd love it if you could talk briefly about what sort of business problems companies can solve with ARC. What are the most common scenarios that we're seeing with the customers?
I mean, lots of money is being invested in AR, right? And so it's being redirected from elsewhere in the business put into this. So, and a lot of competitors are releasing AI. So, how do they stay relevant in the market? How do they make sure that the money we're putting into our products and…
Cool. Well, listen, thank you so much for being on the show. This is a really fun conversation. I hope it's fun for the audience as well because it's really exciting to us. For people who want to learn more about ARC, explore whether it could help them evaluate their own experiences. Where should…