Ep 47 - AI vs. Search Engines: Who Wins the Battle for Information?
Watch the YouTube video version above or listen to the podcast below!
Ep 47 - AI vs. Search Engines: Who Wins the Battle for Information? Podcast and Video Transcript
[Disclaimer: This transcription was written by AI using a tool called Descript, and has not been edited for content.]
Dave Dougherty: [00:00:00] All right, welcome to episode 47 of Enterprising Minds. Alex has the topic today, so I will quickly bring it over to him. You are in luck because Ruthie's here for this episode. So she will provide most of the insights, most of the values, like usual. So, Alex, topic for the day.
Alex Pokorny: You know, before we hit the topic, let me do a little icebreaker question, because I imagine we will have some distinctly different answers to it.
Super Bowl Talk
Alex Pokorny: So the questions are number one, did you know the super bowl is on Sunday? And number two, can you say what time it is kickoff, at least in your local time zone?
Dave, can you answer these questions?
Dave Dougherty: [00:01:00] Yes, but I'm not going to, I'm not going to give it to her because I want to know, I
Ruthi Corcoran: didn't know the Super Bowl is coming up in the near vicinity. I have no idea what time kick out this and I think the Eagles might be involved.
Alex Pokorny: Oh, yeah. Okay. I'll give you half point for that 1 as of this morning or the new service.
I saw an email. I was like, oh. Super Bowl's on Sunday. No way. And then later I googled it so I could figure out what time it was. Which turns out 5. 30 in our time zone.
Ruthi Corcoran: Wow, that's so late. We're going to be in bed like an hour later.
Alex Pokorny: So for the Who are you? Event.
Dave Dougherty: You go to bed at 6? We are not representing well.
Ruthi Corcoran: You know, I put my kids to bed and I call it a day.
Dave Dougherty: Dave is just stunned. [00:02:00] Okay, we're living very different lives. Very different.
Ruthi Corcoran: Good kick off question. Good kick off.
Alex Pokorny: Yeah, I had no idea. It's just hilarious because I used to watch it. This is so bad. I used to run ads for some brands for Super Bowl and unless you're licensed, you know, if you ever watched the game, it says you're the official license potato chip of the Super Bowl or whatever, you're the official licensed gas station of this.
I mean, they literally have one for everything, but if you don't have that, you don't have the contract. Therefore, you have to say. Game day instead of Superbowl. And the ads always were just so lame because there was always like, get ready for game day. And it was like, that's so generic. The
Ruthi Corcoran: big event.
Alex Pokorny: There's, there's a local bank that we
Ruthi Corcoran: get today.
Dave Dougherty: Sorry, Dave. Yeah. Yeah, no, there's a local bank like that where they want to They, [00:03:00] they want to be associated with the NFL team in Minnesota, but they don't pay for it. So then their ads are always, go purple. Like, ooh, just don't, just don't.
Alex Pokorny: Just random colors.
Just randomly they throw up one that's like, go blue. And everybody's like.
Ruthi Corcoran: Here's one for the home team.
Alex Pokorny: Right. Cool. Yeah. Whoever those people are that we cannot name
Dave Dougherty: contract. Yeah. Sometimes you should just pay for the real thing. Yeah. Yeah. Oh, so funny.
Alex Pokorny: All right.
SEO and AI Search Trends
Alex Pokorny: Question actually for the day. I was recently talking to a consultant and was trying to give him some SEO advice.
We're talking about his different clients. So he's got a little variety and we'll bring in the variety later, but we'll start out with his first one, which was a local veterinary clinic. And partially tied to this question is also a little stat that I'm going to throw out here too. [00:04:00] The kind of marketing, but also estimates and number of like different aspects that that company kind of puts together, they call themselves intelligence company.
It's a little generic, but they put together a stat that says by 2026 search volume on all search engines will drop by 25%. And they just put that out just recently. I don't actually know if it was last summer, but recently enough, so 2024, 2025, they said by 2026, they said, real short time frame, gonna drop all over the place by 25% largely attributed basically to the rise of AI search.
And interesting little side note on that is as of like a year, year and a half ago, I mean, chat, GBT was basically hitting like perfect early adopter kind of percentages, meaning like 15 percent of the market had tried it. I mean, there's a whole chasm in the center as we kind of see our distribution curve, just barely getting that first one starting to hit that second one kind of a thing.
I'm sure we're now probably well [00:05:00] into that second one, but do you really. So it's kind of a two parter kind of question of one kind of general SEO advice for small businesses today going into the next couple of years are going to be your times of transition and then two, do you believe this Gartner number do you think that AI search is really going to take down global search search engine traffic?
So
Dave Dougherty: my take on this is I have a hard time believing stats like that is really good headlines and it's good for them to be able to have people go, well, that, you know, that's like 70 percent of my conversions, you know, if I look at my analytics, but what am I going to do, you know, how do we do this? And, you know, we've talked about it before where SEOs aren't just search engine specialists, they're findability specialists.
Right. Like you actually have to figure out how the internet works, how the search [00:06:00] engines make sense of the internet, and then how that technology ultimately is being leveraged by the AI. Models. Now they're all very related, all very connected, you know? So I think the SEO world is in a good spot to handle that transition.
But I think the, the data that we have now, right. Anytime you, you know, if you're data driven, ultimately what you're doing is you're just looking backwards and trying to say, yeah, the future is going to be exactly like the past, which we know is not true. So I think the, the data right now is all of the early adopters is the tech people going, ah, I'm going to, you know, I have to figure this out before everybody else does.
So it's just, it's a, it's a weird snapshot of. The overall market. And I don't necessarily think it's representative because [00:07:00] I was thinking about this the other day too, where, you know, we'd spent a lot of time when Chad, GPT first came out talking about it on this podcast, just trying to make sense of it for ourselves.
Right. Let alone, you know, everybody else. And then I, I thought, man, so many people are going to adopt this and just like, see the writing on the wall and, and like, we're in trouble. Like, I got to figure this out like now. And then. No, nobody did. I mean, they might've messed around with a little bit, but like most normal people that I know, they've heard about it.
They might've tried a free version like here or there if they're curious. But for the most part, they're just not using it right now in the marketing side a lot more people are using it, but even to the extent of what I thought people would be using it for, I'm not seeing it. You know like the wholesale adoption.
I think [00:08:00] Ruthie, what you said last time and the last episode of like, it's actually an investment. You need to keep up on all of the updates on the LLMs to even know what they're capable of now because they keep launching all these new features and new ways of doing things. So it's just hard to know how that's actually going to hit at this point.
Ruthi Corcoran: I'll use that as a good segue into some of the initial thoughts I had. From Alice's question okay. So the 1st. The 1st comment I've got written down here is I was listening to Adam grant. One of his, I think it was either rethinking or work life with one of the founders of Airbnb. Great podcast series fantastic interviews that cover all kinds of things.
And at the end of this particular discussion, Adam Grant had asked, you know, Hey, what's your hot take on Airbnb or on AI? And the response was like, I don't it's not going to change as [00:09:00] fast as everyone expects, but the changes ultimately. Will be more drastic than anyone's anticipating, probably in ways you can't even anticipate.
So I always keep that top of mind of there's likely to be a lot of changes that we don't even know are going to be changes. Until we see there and we're living in a very different world. And so I think that dovetails nicely, Dave, with the comment you made about, hey, JETSY, JETSY came out and we thought like, oh, it's going to be amazing.
We're just going to use it. And it's like, oh. It's kind of taken off, but it's not the mass adoption. They're sort of It's like a slow roll, right? Or I suspect the avid avalanche will catch up and all of a sudden it'll be a very drastic change, but a slow roll.
Personalized Search Experiences
Ruthi Corcoran: In terms of, man, I'm thinking about this SEO statistic of like search going down by 25%.
And the experience that comes to mind is something that I wrote, I wrote down some notes a [00:10:00] couple days ago. I was perusing the web, the internet, as one does, for flowers. We are in Dahlia tuber purchasing season. If any is, anyone has ever heard of tulip mania in the Netherlands a few hundred years ago.
People like to claim that dahlia mania is a thing now, where people just buy a lot of dahlia tubers and they spend boo boo bucks on it. Well anyways, it's sort of a phenomenon that happens in certain parts of the internet around January when all the tuber sales go online. So I'm taking a look, I'm looking for, hey, what am I going to be purchasing this year for all of my pots?
I've got a couple of go to websites that I, that I take a look at to see, hey, like what are their, their offerings? And what's amazing is as I search for various flowers or plant varieties, boy, the Google results are rough. They're just. It's the same big box retailers over and over again, not just in the paid, [00:11:00] but also in the organic results.
And
it's really annoying, frankly, because I don't, I don't, I, I already saw what they have. I want somebody niche who has a different variety or very specific plant. And it's very hard to get search results to display different things than sort of what the big box common search results are. And I'd like to say I'm pretty decent at making search engines respond with the types of things that I want.
And it's really hard to get them to offer you tailored results or something that's a little distinct. It has to be very long tail and even in that case, it's very difficult. So I'll give you an example. I checked out a couple of sites. I happened to find this particular type of Dahlia called a bloomquist cracker, like this really cool red flower.
And I was like, that's amazing. I want to find a seller who not only has that [00:12:00] variety, but five other varieties I want to buy at the same time. And so I searched for that. Same result, big box retailers, not even selling that particular tuber. It's showing me a bunch of other places that sell Dahlia tubers, but not that particular one where I'm going with this is ultimately, I hit a website that I've used before called dahliaaddicts.
com. And what it does is it lists every single variety and it lists all the different. Places that sell it with their price. Presumably they work with sellers or sellers can list their variety on the website. And what's amazing is how old school is that? I'm going to this one particular website that is sort of the hub for all these different things.
I can't go to Google to do that. I have to go to these very niche websites. And the reason I bring this up is because, boy, if that's the current experience. And AI can do a better job or different AI models can do a better job of [00:13:00] finding this kind of niche information and serving it up in a personalized manner.
To me, of course, mainstream search is going back down by 25%. Of course it is because mainstream search isn't personalized. So I wonder if it's something more. That these various different AI models are going to take the role of our current search experience over time to be more niche, to be more personalized.
I can give more specific information and I don't just end up with the same five websites over and over and over again.
Dave Dougherty: Yeah, no, I think the, the personalized pieces is an interesting direction of that, just because as you were talking, I started thinking like, okay, well, what. What are the search engines using or like what was promised us for good results?
And so much of it was based on your search history your click activities, all of these things that are now barred by The data [00:14:00] privacy laws, you know, either in particular states or globally. So then a natural reaction to that would be, okay, well, then what's a good proxy for generally good information, an established brand, right?
So it's like, I understand that logic, but to your point, if you are into a particular thing, it's really hard to find those because. Yeah you either, you either go to the company that you know, like for me with guitars, I have narrowed down my interest in guitar builders to two or three. One of which does not have very good distribution because it's a mom and pop company from like 1950 that's in California and it's still family run and they still hand make them in California.
So you have to pay out the wazoo for [00:15:00] the product. Oh my God. Is it like the best thing that you've ever played? Right? Now it has taken me 20 years of playing to get to that point. But now that I'm there, I really can't go back. So, you know, you have to sign up for other things like the newsletters, like the, you know, alert the company alerts or, or whatever to be able to get the information that you need instead of relying on the guitar center.
com or sweetwater. com or, you know, whoever else.
What did you want out?
Alex Pokorny: Where are you going to take it? I was just, it was a bit of a mind blown statement just there of, I think I finally understand what AI could do that Google can't do. So we talked last time about how really there is no cloud. And in that [00:16:00] aspect of when you search the internet, I was just talking to a friend and kind of, it's kind of, it was all around this, but I was like, when you talk about you search the internet, you searched Google's index.
If you didn't search the internet, that's two different things. The internet, to be honest, is more of a concept than anything, but it's all of the websites and all of the web content out there. But that's not what you can access on Google. What you can access on Google is just what Google has indexed. But then on top of that, it's what's Google's algorithm shows you out of that index.
So it's a bubble within a bubble kind of thing, right? And if you're stuck and you recognize the edges of this bubble and you're dissatisfied with it, you don't really have a great alternative. Microsoft, yes, is a slight alternative with Bing, but if you look at search volume share, no. And if people are really, really disappointed with something, they're going to basically eventually change behavior.
Or if they find that, hey, somebody found a better thing. It could be better deals [00:17:00] on flights. It could be better, you know, something else that Google isn't surfacing to you and someone else is, you know, saving money or making money or something like that, that will definitely cause a drive in that direction.
And if AI can crawl. Enough A. I. Companies can crawl enough sites to build enough off of an index that has enough variety to it. You can say, Hey, on perplexity, I found this deal for this flight. But on chat, I found this deal for this flight. And Google, of course, was two times that amount. You're going to see adoption radically change up.
Dave Dougherty: Alex, it's interesting you bring that because here's at least my first thought in response to what you said. You know, just like we've talked with some of the Clayton Christensen management principles, the very thing that is going to drive people to AI will be the very thing that will be its downfall as well.
Because just like the search engines as they are now, it requires a certain amount of personal [00:18:00] data in order for you to leverage all of the features. And it's up to the individual to decide how much of the data, how much of my personal information am I willing to share in order for this to actually be feasible and work, right?
For example, the operator thing that has come out with ChatGPT or some of these other ones where it actually clicks and does things for you. Like, do I love the idea of just saying I want to, you know, I want a family vacation to Greece here. Our favorite islands set it up for this weekend. Yeah, that would be amazing.
Am I about to give a non sentient being my credit card? No, because I have no recourse if it screws up, at least if I have a personal assistant, I can sue them and fire up. Right. But if, if, if I screwed up because I set up the bot wrong, that's on me, you know [00:19:00] and I just, I, for me personally, I won't, I will, don't imagine myself being in a position where I'm comfortable with that amount of.
My information out there, right?
Ruthi Corcoran: I also want to go back to this idea. You talked about the surface of the bubble and Man that's such a cool way of describing it and this idea that when you're searching Google, you're not searching the Internet It's true. You're searching Google's index and The things that that come to mind are hey, you know The wealth of knowledge available on the Internet has changed our approach You know, it's changed our options that and it's changed our expectations about what's possible.
And about what we can, you know, possibly get either via information or product or connection to other people. It's just changed our expectations in a really big way. And I think a [00:20:00] couple of things 1st, it's not so much that search is getting worse. It's so it's more that our expectations have expanded beyond it.
And being able to match those new expectations while also trying to manage the competition for attention within the marketplace of Google's index is very difficult. And I think we've had a couple of conversations about this, where people are spending more of their time and attention, and they're They're getting information from platforms like Instagram from platforms like TikTok or niche listing sites like Dahlia Addicts.
And I think that's the reaction to hitting that edge of the bubble that you talked about and they're trying to find an alternative and right now it's in those little niche places. And then to Dave, some of the things you talked about, right, the, the value of AI is it's going to allow you to better access those niche areas that are beyond the current bubble of the search engine.[00:21:00]
That's cool. So if I'm a local SEO person that's curious, like, that's kind of cool in a couple of ways. So one is like small business SEO. I think one of the Right? Maps is where it's at being able to search a random thing and having maps serve up. Oh, we think that this company or this organization has the thing that you're looking for.
And it's right down the street. That's really cool. And I, I don't think that's going anywhere because people use maps. They're going to continue to use maps, whether it be apple maps, google maps, some other map like that basic mechanism, I think is going to continue. I think what shifts though, is if you're a small business looking to.
Reach a wider audience, especially that niche audience. That's on the other side of the country. Now, all of a sudden, you potentially have a new outlet and you don't maybe there's a way in which you don't have to compete with the big boxes. You're able to get that niche audience. I'm not exactly sure what that looks like.[00:22:00]
Is it you double down on the Dahlia addicts dot com type sites. You make sure you're available on those places so that a given AI assistant can access it and sort of has the connections. Maybe that's the way to go. I don't know if that's a question I'm, I'm curious your thoughts on.
Dave Dougherty: I do think it depends on your model.
Local SEO Strategies
Dave Dougherty: I know in a couple of our in person off camera lunches, you know, we've discussed different things and, I had met with a former co worker who's now doing real estate and Just to bounce ideas off on, you know, what's the potential strategy. Cause you know, like it's one thing to be a marketer.
It's another thing to be a business owner who is marketing, right. That's a very different thing. Cause you have more on your plate. And he asked me about SEO and kind of the digital marketing mix and all that. And I said, like, you don't really want SEO when you're starting out [00:23:00] because it's a mass communication tool.
Like, and you're very focused on this particular part of the twin cities and. That's great. If you want to be known as, you know, that guy in that part of the, of the state, that's fantastic. But that means you set up your Google business. You make sure that you get people to leave you reviews, you have your website so you can maintain an owned media experience so that you can control your messaging, you know, for whatever happens.
And all of your other marketing should be within that locale to build up your name within that locale. So, sponsoring the. The elementary school choir or you know, the local dogs with cancer, five K or, you know, whatever the charity event is that you can get involved in for however much your budget is, right?
I mean, all those are [00:24:00] gonna, but then you show up and you participate in those things too, so that they can associate your ad with your face. In the community, right? And that's a very old school tactics, which again is like anytime the technology changes, what's old is new again. Because I think that's really what's going to separate out the people who are serious and the people who are, you know, in their basement or in their apartment who don't want to talk to people and just use the AI to get a portion of the market who just want the information and they'll try it.
And if they don't like it, they'll throw it away. I mean, that's, that's definitely a section of the market, but. It's not where I like to play, you know so my strategy is going to be different.
Alex Pokorny: I think I had to kind of since I'm reflecting on a conversation, I've kind of had a there's a mix of both of yours. So from a content perspective to Ruthie's point, I mentioned [00:25:00] since this is a veterinary clinic if they're bringing in, let's say a surgeon to do additional surgeries or something like that, list those services, but make sure you list basically everything that you.
Can do on your site, there's different types of vets out there as well. Some only do, you know, small animals, some do cats and dogs. There's a whole bunch of different types as well. One of the other ones was also technical SEO as table sticks, basically using any of the free tools that are available online to basically kind of.
Go to a decent review of your site just to make sure that it is crawlable and indexable just kind of like Basic things if you built your site probably with WordPress Squarespace something else you're probably fine If you built it yourself, I would definitely be looking at tools and looking very closely at them More so than if you used a popular Popular builder.
And then yeah, Google Maps basically was the other one was basically saying like that is a far easier algorithm to rank in than trying to rank in kind of the general bucket of everybody else. And plus you don't want that national traffic anyways. So the main thing was basically after that was looking at [00:26:00] Google Maps in terms of a radius.
So if you're in Europe, this doesn't apply as well, but in the United States, it's about like minutes away. So if you're 20 minutes away, that's your Distance instead of kilometers or, you know, off sun, you know, public transport routes or whatever it might be that might be more convenient to your local clientele.
In this case, it would be a driving clientele. So minutes away from the house looking at that radius and saying, this is my competition set and how do I beat this competition set? So looking at their sites, going through their information and trying to understand what's different between all of them.
There's an old advertising trick really, but it's basically if you're, if you cover up the name of a brand, your brand with your thumb and then read the slogan and you replace it with a competitor. If it's still true, it's a terrible slogan. So it's like, you know, whatever that in whatever city since.
1950. If there's a, I mean, that doesn't really mean anything or you provide quality care what your competitors don't like. A lot [00:27:00] of these statements just don't mean anything. So trying to instead remove the generic frivolous kind of statements instead trying to focus in on. What makes you interesting?
What makes you different? Highlight those items, but make sure, at least from just a general search perspective and AI perspective, your site is crawlable, indexable, and then lists all the things that you can do. I also showed him also asked. com to show kind of different queries that you could if you want to build out FAQ content or kind of brought it out a little bit.
But kind of note that if you're trying to be a vet clinic, stick around with vet clinic. Don't go to dive deep into we saw a bunch of queries on vet billing. You're going to get national traffic. That's really not useful to you. If you're trying to get leads, talk about the local area, talk about what you do and try to get that keyword rich kind of information up there.
So
Ruthi Corcoran: I think some of the thoughts that come to mind too, as you're describing this vet clinic works well with what Dave said, trained your local vet. Clinic. Hey, find the local dog parks. Oh, yeah. [00:28:00] You know be the be the cleanup crew and put your little sign up there, right? There's something about Building community and tapping into the local community that is so important as part of a local strategy, because to your point, it is a minutes sort of situation.
And if they're seeing you within their Google reviews. And it's not just that you have the service that somebody is looking for, it's, they also recognize your name and they see your name and they go, Oh, that's trusted. That's fantastic. Like that, that plays into it greatly. And I, gosh, I, one of the things I think about with this.
The rise in various AIs is, and we've talked about it here on the show before, is that trust factor. And this just keeps coming back [00:29:00] and back because if you're, if the copy that you see and the videos that you watch could all be, could all be created by an AI and there's not sort of a human connection.
You got to find different ways of establishing trust with potential customers that doesn't rely on digital content.
Alex Pokorny: I think showing your face is important. That was just a random comment for the small business side of trying to not just be a name. Because building a brand, especially these days, is quick and easy. I can throw off a brand and a site in the next hour
Ruthi Corcoran: with beautiful stock photography.
Alex Pokorny: Exactly. It doesn't really mean anything.
And I can get listed on Google with generic photos of a generic place or just, you know, random logo imagery, but that doesn't mean anything. Again, that connection point, the trust point [00:30:00] like that has to be built and it's still a human connection.
Ruthi Corcoran: The the other thing that's, it's related to this conversation, it's something I've been noticing a lot recently is the frustration I have with FAQs or reviews not answering the question that I care about.
Oftentimes, for specific products, there's things you want to know. How well is that shirt going to wash the seventh time? Right. And it is so hard to get the answer to that.
Yeah.
It's amazing. Like there's so many, that's a, that's another angle I think is sort of what are the content niches related to your product that are just simple and easy to fill?
And I'm not saying go buy fake reviews to say, Oh, I watched it seven times by any means, but okay. Put it in your FAQs. Hey. This is what you can expect in terms of level of [00:31:00] shrinkage or whatever it is that is the problem for your particular product that people commonly run into.
Alex Pokorny: You could also like put out a, just speaking as a brand standpoint, you could also put out as an e commerce platform as well saying that.
Products from this brand typically wash well or something like that. You can put up some statements like that saying that in other reviews of other products, because it might be a brand new product, you know, might not have that many reviews yet. You can stay those kinds of statements of saying, generally speaking, this brand runs small you know, whatever else, kind of like trying to give you something, because I think one of the big annoyances I have commonly with reviews is that the person, the purchaser was asked to give a review too quickly.
And what I get is. Packaging. Yes.
Ruthi Corcoran: Yes.
Alex Pokorny: It arrived. It looks okay. And it's like, but does it work? Oh, we haven't tried it yet. Clearly. Yes. It arrived dented. It missed some parts. It's broken. One run rival. The package was damaged. You get these kind of queries, these like very basic things of saying, it looks nice.[00:32:00]
I'm sure it will work fine. Five out of five. And it's like, so you didn't try it yet. Five out of five
The Future of AI in Search
Ruthi Corcoran: and where this circles back with some of the, the, the future of search and AI conversation is that right now you can try and get that niche within your search query. But it's pretty raw. LLMs make it much easier to articulate.
Here's what I'm looking for in a product.
Yeah.
And if potentially they have the data in which they can say, Oh yeah, this particular product meets this thing that we understand you're trying to ask for. Maybe not understand isn't the right word, but Do you see what I'm going with that Alex?
Alex Pokorny: Yeah, absolutely.
I do. This was a recent personal example. So I was trying to find it was a potassium supplement. So I was searching online. It was a random thing, but I was looking online and I was trying to figure out, and I found on Amazon lots of different amounts [00:33:00] that I can purchase. And some people are taking them multiple times a day.
Okay, you, you have too much of this. It's not a good thing. So what, what, what's the amount? And I kept getting NIH the UK health organizations. Required amounts for men or something like that for a daily consumption. I'm like, but this is a supplement. This doesn't mean my entire diet. So assuming, I don't know, an American diet, how much of this do I need before I am taking too much?
Like, is there an assumption that my diet provides is X amount? And then that's the amount. Am I supposed to, and then some are saying this multiple times a day. Like, it was so funny because this information was so disjointed there was the quantity. On the product, there was a daily amount, you know, suggested, but there was no time between that said, this is the amount that I should be taking.
Ruthi Corcoran: Yeah,
Alex Pokorny: and that was the piece that was frustrating to me. It was, this is, [00:34:00] it just smacked me in the head of like, this is the current search experience to a tee. I get this e commerce result. I get this generic top query and I get this middle that's missing that basically that's where I see where any AI tool could pop in and say, and it says this with that a little bit of reasoning model, your diet likely provides half of that.
You don't want to take too much. So you should take a most 1 of these milligrams. Here's 3 products that you could buy that would. I don't fit that amount. Something like that would be much more sensible. And there's also like other added information that I don't have as well which I think is always the narrow aspect of Google, where if I am taking this, should I also be taking something else?
Or should I just be changing my diet a bit and involving certain types of fruits and vegetables that would be easy for me to consume? And I don't even need this whole thing. Like there's these added elements. I'm, and it's so hits me with this idea that we are still searching [00:35:00] keywords and getting keywords back.
The difference is if I talk to my doctor, I can say, how much do I need? He will tell me, he'll tell me the amount. He'll give me warnings. He'll give me, by the way, you should also be doing this or Hey, by the way, you could just forget all of this. You need more. Bananas as it turns out are not enough, but eat something else and that would that would be fine Like it's the complete answer the expert answer and that's the piece i'm missing like the the dahlia question of i'm looking for these items But i'm looking for them of course packaged together so I can get them shipped at the same time And I don't want to work with five different companies and go through the shopping experience five times when I can just go through it once and I have
Ruthi Corcoran: I I want to uncover more.
I want to discover more like this And if I know this,
Alex Pokorny: what else should I know about it? Or you should tell me also, hey, by the way, this one looks really cool and it seems just like this other one, but it turns out it is highly temperamental and it won't work for your particular situation. Like, those kind of pieces which a [00:36:00] friend who is a neighbor would tell you, an expert.
Someone who's an expert would give you the full answer the complete answer. I think that that's that's the piece we're missing the frustration I have of course with any AI tool is the moment they become experts I start questioning it and start finding holes where it's hallucinated very highly questionable There's a whole bunch of URLs.
Oh, wait, those URLs don't contain any of the cited information that apparently has been cited by these URLs. Like, oh my gosh, I was doing one yesterday. It was so frustrating with ChachiBD and it kept giving me diagrams that were literally the exact same image. And it's like, do you want a diagram? I'm like, yes, this is a, an angles and design kind of question.
It was like, yeah, give me a diagram. And it shows me the diagram. Here's options two and three. It literally gives me the exact same image two more times. I was like, come on. It was like, Oh, my mistake. And it gives me, of course, another set of three images that are exactly identical. I'm like,
Ruthi Corcoran: And [00:37:00] notice this idea that you brought up of like the missing middle, this sort of delta between the current keyword search experience and what we, we, we think AI will be capable of doing.
Notice our assumption is that the information we're looking for exists. And I suspect for a lot of the topics we've talked about, it does exist, right? We've given examples where it's like, yeah, you're Your doctor or your average, you know, gardener down the street would have this information. And I wonder too, if once, once these AI becomes ubiquitous and the information that's out in the world is accessible in a much more particular contextual way, once that becomes ubiquitous, we're going to reach a new bubble, which is the actual limits of human knowledge.
And then to your point, part of the trick is going to be, [00:38:00] is, right, are the, are the AIs we're working with good enough to know, like, this is the limit of, of our knowledge versus like, this is speculation versus I just made this up? We're going to start encountering that and then our frustration is going to be we haven't figured this out yet Why hasn't somebody studied the very particular growing habits of this particular variety of dahlias in Minnesota in my particular soil?
And then we're really going to be annoyed
Alex Pokorny: Yeah That's going to be the edges, but that's that's the edges of an expert too because you can ask them that same question They might be an expert. I was watching this is last summer some gardening videos from Canada and Similar enough climate, I thought I was like, you know, this should work. So it was fantastic to hear from them on what, there was a gardening clinic and they had a bunch of interviews and they're going through like, here's some, you know, you know, idiot proof gardening things, which I was like, yes, this is what I'm looking for.
So it was great content for me. [00:39:00] But it was able, I was able to then take that and expand that because it said this is what's good. And there was a, there was a local organization. So it was meant meaning their area, but I was able to then with enough knowledge of zones to understand, yep, the growing zone is a little different for me, but it's going to be good enough and it's going to be close enough.
So able to expand upon that with. Basic reasoning, which I think is also like a capability that I can't have is that extra edge, which says probably this will work instead of these confident answers that I'm getting today from different AI tools is basically like, this is going to work. And it's like.
You're literally saying something that's not even true. Like that's, that's at least, you know, maybe color code. Your answers are saying these things and blue are solid and these things in red are questionable. The skill of the AI
Ruthi Corcoran: age. Bullshit detection. Exactly.
Alex Pokorny: Like give me an edge on this of like, because that's the thing is that expert tells you those things of being like it.[00:40:00]
Yeah, it probably will be okay with your environment, but it's really just meant for this environment. And then you take that information, you say, okay, I made this person pretty uncomfortable with this question. I'm giving this maybe a 40 percent chance of working and maybe it's still worth it to me.
I'll still give it a shot. Why not? Or maybe it's not, but an expert can expand upon that. And yeah, humans are much better to Dave's earlier point of looking at past information, past history, and being able to try to think that trend line is going to continue in the, as it goes forward. And people are terrible at making estimates.
We know this, I mean, there's no scientific proof of that, that people are terrible at, I mean, heck if they did, that everybody would win the lottery. It's just, we're not good at that.
Ruthi Corcoran: Do you know what would be really cool though? Is he had talked about sort of the expert sort of we hope most of the time, [00:41:00] right?
When you go to the doctor, they say, hey, this is this is sort of what we know. Here's your range of options that are you get to make a decision. That's the best case scenario. Sometimes you get you get. Experts who just say, yeah, this is the way to go. And it's total nonsense, right? So that happens in the human world as much as the AI world.
What I was thinking about though is there's a, an economist, her name is Emily Oster and she has become much more visible in the public eye because she wrote a wonderful book called expecting better. And now has a podcast and a whole content series. That's. That's purposes. Let's give people the data that we have so that they can make decisions for themselves.
And so expecting better is all about like, Hey, here's the thing is we actually know what pregnancy and whether you should eat sushi so that you can make a decision. And how cool would it be if that was one of the models in which we engage with, with various AI is what, with the information they provide is, Hey, here's the information we know.
Sort of laying it out in the way that you make the decision now, [00:42:00] as, as the human actor interacting with the AI,
Alex Pokorny: that'd be wonderful because that's the research that we do on Google right now. Google has the shotgun idea of, I don't know, here's 10 different pages that might have the answer you're looking for.
Well, I mean, it's infinite school now, so it's even more than that, but. That's their response to it versus giving you a complete, you know, answer, kind of the librarian answer versus the library answer. That's what I really want. I want the librarian to tell me what the kind of the full is, instead of just giving me a stack of books for me to read myself.
Conclusion and Final Thoughts
Alex Pokorny: So, well, this is a good place, I think, to close it out. Ruthie, fantastic conversation. Glad to have you a part of it. Dave as well. Thank you all and see you all in the next episode.
Ruthi Corcoran: [00:43:00] Cheers.
