View Auto-Generated Transcript
The below transcript has been auto-generated for your convenience. Please reference the source video/audio for direct quotes or to clarify any errors.
Well, one last question I want to ask you before, before we run out of time, which is something you talked about, you mentioned both two terms that I, one, which we hear a lot and I. Can't decide if I think is brilliant or cringy, which is prompt engineering.
And the other, which I actually think is a novel, more interesting is AI literacy. And where would you put when you're thinking about hiring, working somewhere, doing something like. Is, I don't know, maybe define AI literacy and also how important do you think that is now and will become
Nate Roybal: I think it's super important. I think AI literacy is a spectrum obviously. And it's between like, Hey, do I know what an AI is? And do I know how to write a very complex prompt or a series of prompts? Right? And so like, yeah, literacy is probably like, hey, do I understand what an AI can do task wise and do I understand how to get it to do that task for [00:38:00] me effectively on a regular basis, right?
That's probably AI literacy in my mind. And yeah, prompt engineering I think is probably going to go away. But with the caveat that I think most people are not used to thinking the way computers think, right? I'm much closer to that because I've been at Mercado and Syncari where I have to do decision trees all day and think about that kind of stuff.
Right. Most people don't think like that. So they're like, why is the AI not working? It's like, because of what you told
it,
Connor: doesn't it, only knows those things a
Nate Roybal: does exactly what you tell it to do. Right. And if you did not do a good job describing That's what happens. Right?
Connor: A connection I just made that feels inspired by what you just said is like, oh, and I think of AI literacy. I think about it as this, a low bar, but I think what you, the comparison you just made, right? Is like, it's not that different of a framework or way of thinking, then like solutions architecture of what is technology capable of and what do APIs do and how can I connect them?
And that's not a, it's a [00:39:00] technical skill set, but it's not an engineering or development skill set. And I think, I don't know. I'm wondering is my, am I wrong about either how hard or how valuable AI literacy is because I'm, cause when I think of solutions architecture, I think about it as somewhat like higher on the difficulty chain and higher on the value chain.
Whereas I think of AI literacy is like higher on the value chain, but lower on the difficulty chain. And maybe that's just wrong. Cause I'm coming from a perspective of previously existing knowledge.
Nate Roybal: I think it's also depending on what you're doing, right? I mean, like some of the process I'm doing yesterday, I'm using like two different models and you know, like 15 minutes across like three or four different prompts that I spent a bunch of time working on the prompts for, right? It's not like it's simple,
right? But at the same time, it's not like it's a writing code. So yeah, I think data literacy and data literacy, and that's a term we use all the time, sorry. AI literacy is is, yeah, it's literally just like being able to understand what's possible too. Because like, there's a lot of times. People go try to use AI and then they're like, why is it not working?
And it's literally because like, they don't, you know, whether they don't have the grit to [00:40:00] like keep trying different ways until they're like, Oh wait, I got the result. Or they just don't understand that. Hey, like there's nuances, right. And they give up, right. Or they just think it's a toy,
right? I think there's tons of people I talked to that are like, Oh yeah, it's still nascent. I'm like, cool. I'm probably saving like 10, 15 hours a week from my team at least. You know, like half a head count. That's great. We're not a big company, right?
Connor: Yeah. Significant. Yeah. What last thing for you? And then I can let you go, which is for anyone who is, has listened to this, is interested and excited about accessibility for some of the custom GPT stuff. Where should they start?