Latest / The Adults In The AI Room / Search Optimization, Competitive Advantage, and Balancing Privacy in an AI-Powered Future - Marc Pickren - The Earley AI Podcast with Seth Earley - Episode #040

Transcript

Search Optimization, Competitive Advantage, and Balancing Privacy in an AI-Powered Future - Marc Pickren - The Earley AI Podcast with Seth Earley - Episode #040 · The Adults In The AI Room

• 0:00
  SPEAKER_00: This is the Early AI Podcast, hosted by Seth Early and Chris Featherstone.
• 0:05
  SPEAKER_00: Join us as we delve deep into the passions, expertise, and experiences of thought leaders and practitioners to talk about what's possible with artificial intelligence.
• 0:14
  SPEAKER_00: The Early AI Podcast is sponsored by Early Information Science, your digital transformation journey with design and deployment of innovative technology solutions, as well as CMS Wire.
• 0:24
  SPEAKER_00: Now, enjoy the show.
• 0:30
  SPEAKER_01: Great.
• 0:30
  SPEAKER_01: Well, welcome to the Early AI Podcast.
• 0:33
  SPEAKER_01: I'm Seth Early.
• 0:34
  SPEAKER_01: Chris Featherstone couldn't make it today, but I'm really excited to have our guest today.
• 0:39
  SPEAKER_01: And we're going to be talking about some fantastic topics, including how AI is revolutionizing the search experience and how it's changing that search experience and what organizations need to do.
• 0:50
  SPEAKER_01: Many of the things the traditional industries around SEO are going to change, right?
• 0:55
  SPEAKER_01: And we need to understand what the implication is.
• 0:57
  SPEAKER_01: We'll talk a little bit about how, you know, uh the big players, such as Google, have so much data that they can leverage to create personalized experiences.
• 1:06
  SPEAKER_01: But what does that really mean for the organization?
• 1:09
  SPEAKER_01: What does that really mean for how we're going to be going to market and getting the word out?
• 1:16
  SPEAKER_01: Talk more about autonomous agents and the implications and the balance between technology and human judgment.
• 1:24
  SPEAKER_01: And there's also the idea of career paths for people, right?
• 1:28
  SPEAKER_01: We know there's going to be a lot of disruption in this marketplace and society.
• 1:33
  SPEAKER_01: And how do we prepare people for that future?
• 1:36
  SPEAKER_01: So our guest today is currently the president of NextNet Media.
• 1:41
  SPEAKER_01: He has 25 years, over 25 years of experience as an entrepreneur and business leader.
• 1:46
  SPEAKER_01: He has expertise in technology companies and demonstrating a track record of building and managing successful ventures.
• 1:54
  SPEAKER_01: He has experience across multiple industries, including fintech, digital marketing, software as a solution.
• 2:01
  SPEAKER_01: He's overseen$100 million PLs at large public corporations and Madison Avenue agencies.
• 2:08
  SPEAKER_01: And he's remained at the forefront of this very dynamic digital landscape and delivering solutions for both customers and businesses.
• 2:18
  SPEAKER_01: He has extensive knowledge and experience in AI and automation and more.
• 2:22
  SPEAKER_01: Welcome Mark Pickren to the show.
• 2:25
  SPEAKER_03: Hey, it's great to be here.
• 2:26
  SPEAKER_03: Good to see you, Seth.
• 2:28
  SPEAKER_01: Nice to see you.
• 2:29
  SPEAKER_01: Nice to see you.
• 2:30
  SPEAKER_01: So, you know, we talked a little bit about this when we were preparing.
• 2:34
  SPEAKER_01: You know, we like to start with some misconceptions in the industry.
• 2:38
  SPEAKER_01: So give me your sense of, you know, when you when you start to look at AI, when you start to look at the changes in society, when you start to look at uh career paths.
• 2:46
  SPEAKER_01: I mean, again, we we're going to cover a broad sweep of topics, but pick one of those areas and tell me what the biggest misconceptions are that you're seeing in the marketplace that where executives have either a lack of understanding or misinterpreting things or just downright uh you know misconceptions.
• 3:05
  SPEAKER_03: Yeah, I mean, I think it's it's easy to kind of focus more on the top level.
• 3:12
  SPEAKER_03: And you know, we could certainly go into a lot of conversation around uh you know, still seeing comments like, not sure about this AI, when uh it's basically the genie is out of the bottle and has been out of the bottle.
• 3:27
  SPEAKER_03: And typically markets lag behind uh where where business and the world is really at by nine to twelve months, it's kind of what we saw with uh uh interest rates and the cost of capital and and people being unprepared for uh for that.
• 3:43
  SPEAKER_03: Um, but that the the end of free money had happened much earlier on in the process and just happened to uh arrive um at our doorsteps.
• 3:51
  SPEAKER_03: But I I like to look at things uh more from uh the employee perspective and and team members, because I think that that's where most of the angst is, most of the anxiety, and we already have enough anxiety in the world.
• 4:07
  SPEAKER_03: Uh, and then you know, where where is their role going to be in in terms of organizations and and where is their career pathing?
• 4:16
  SPEAKER_03: And you know, we have uh at our company, we own a bunch of subsidiary companies.
• 4:22
  SPEAKER_03: One of the areas that has been probably changed the most by AI is um is in content creation, right?
• 4:31
  SPEAKER_03: We produce about 32,000 original content or search engine optimized content pieces per month.
• 4:38
  SPEAKER_03: And in the past, we had teams of you know uh 350 contract writers uh you know scattered throughout the globe and here in the United States.
• 4:48
  SPEAKER_03: And the way that um, you know, writers in particular are very a lot of anxiety around AI because when ChatGPT opened up as a free service, you know, basically anybody can go in there with uh with a very comprehensive prompt and and create a piece of content.
• 5:06
  SPEAKER_03: And um, and so what we've been working with both with our writers as well as is as you know, our our team members, our our young managers, is really in looking at AI in terms of how can you be the human in the loop?
• 5:22
  SPEAKER_03: How can you be the responsible party for those autonomous agents?
• 5:27
  SPEAKER_03: How can you develop a skill set of being able to manage autonomous agents?
• 5:33
  SPEAKER_03: And then for the content, if it's an autonomous writing agent, then leveraging the technology to do the early stage research, to do all of the different things that set you up for success.
• 5:44
  SPEAKER_03: But then ultimately, as the changes that you alluded to in terms of Google search, search generative experience being, then you can you can actually you're probably going to be more important to the process, but it's going to make you far more efficient.
• 5:58
  SPEAKER_03: Now, the biggest question is how do you write set what is the going rate for um human beings in terms of their participation in the workforce when you have autonomous agents that can do a lot of the more you know menial uh repetitive tasks.
• 6:15
  SPEAKER_03: And so we just encourage, you know, we're looking at the the fact that we're going to provide for our team members career pathing for them to learn how to manage uh and leverage AI and create probably very um marketable skill sets for the for the for the for the near future.
• 6:34
  SPEAKER_01: Interesting.
• 6:35
  SPEAKER_01: And you know, you talk about um how the fact that you don't need um all of the oceans of of humans, you know, the mass, the hordes of humans to create the content, but uh, but we also talked a little bit about uh uh what happens when we leave this too much to the uh to the systems, to the large language models.
• 6:56
  SPEAKER_01: And I'm assuming when you say autonomous agent, you're referring broadly to uh some of the chat GPT types of applications and large language models.
• 7:05
  SPEAKER_01: Um uh but there's other implications around what autonomous agents, because they can have even more autonomy when we connect them to other things, right?
• 7:13
  SPEAKER_01: Which is kind of scary.
• 7:14
  SPEAKER_01: Um but uh but talk a little bit about what happens when it's almost like this snake eating its tail, because if you are creating all of this uh machine-generated content, um, you know, I've heard of researchers talk about model collapse, where you know, there's just going to be this flood of content on the internet that is generated by large language models.
• 7:38
  SPEAKER_01: And the large language models continue to feed on that, to continue training.
• 7:43
  SPEAKER_01: And then you really start losing that human creativity and you start losing like you can't sustain forever.
• 7:50
  SPEAKER_01: So, do you want to talk a little bit about that?
• 7:52
  SPEAKER_01: And I understand that's where you're talking about the right mixture of human in the loop because you still need to add that inject that creativity into the system, right?
• 8:00
  SPEAKER_01: To keep it from collapsing.
• 8:02
  SPEAKER_01: But do you want to talk a little bit more about that?
• 8:04
  SPEAKER_03: Yeah, I was uh I was I I forgot the quote, but it there was actually a quote on one of your previous uh podcasts that was talking, I think with Stephen Hawking that was you know talking about every contribution to civilization, you know, came from human intelligence.
• 8:17
  SPEAKER_03: And so and you think of that in terms of contribution, right?
• 8:21
  SPEAKER_03: Like value.
• 8:22
  SPEAKER_03: And I I think that's the that's kind of a line item um in terms of like if you view if you view content, let's just say, uh from a from a quality context, well, you know, do is it grammatically correct?
• 8:38
  SPEAKER_03: Does it have the right topic as far as search engine optimization?
• 8:42
  SPEAKER_03: Is it uh uh written towards the keyword?
• 8:46
  SPEAKER_03: And is it it is it technically framed in the right way to be acceptable to Google's algorithm and to and to be you know into the asset uh to be indexed?
• 8:56
  SPEAKER_03: But then there's this like value, like is it providing value to Google, its users, and um and our clients or or you know, clients out there businesses?
• 9:08
  SPEAKER_03: And you know, I'm kind of guilty.
• 9:10
  SPEAKER_03: Uh, you know, when I when I became president of Next Net Media, um, you know, I was I was like very much uh uh gung ho in terms of saying how far can we go with the um with the AI?
• 9:26
  SPEAKER_03: And when I say autonomous agent, like you can create through prompts an autonomous writing agent, right?
• 9:33
  SPEAKER_03: But the the value of being able to provide value to a particular audience and to connect on a human way, in a human way is something that doesn't really, it isn't really possible through um uh through AI.
• 9:49
  SPEAKER_03: And so I pushed the team really hard, and then it it kind of came back to like this is about as far as it can go right now before it starts to just become uh a part of the virus that's going into the search engines and into these places.
• 10:06
  SPEAKER_03: And so um it was a really good experience because there was the human beings in the company that were, you know, that that came to that conclusion, but it was also qualitatively uh it was always also qualitative or quantitatively proven in terms of uh uh you know it not being good enough if it's not providing value.
• 10:27
  SPEAKER_03: And the other thing I would just say is, you know, when AI goes wrong, it goes wrong in in very uh uh uh very uh mysterious ways and unpredictable ways, yeah.
• 10:38
  SPEAKER_03: Unpredictable ways.
• 10:39
  SPEAKER_03: And so when I talk about the employees and staff members and professionals, I'm talking about you know, a human has to be responsible for what is the output of the autonomous agent or the whatever.
• 10:54
  SPEAKER_03: And you're right, if there isn't that human in the loop, then it you know becomes kind of a it just becomes a race to the bottom.
• 11:03
  SPEAKER_03: And and and that bottom um you know will eventually make that it, you know, will create that there's more work for the AI than it's than it's worth, um, if potentially.
• 11:15
  SPEAKER_03: So that's that's kind of how I see that for sure.
• 11:18
  SPEAKER_01: Yeah, no, it makes a lot of sense, and I I completely agree with you.
• 11:22
  SPEAKER_01: And and you know, when I use Chat GPT, I I use it to generate ideas, maybe an outline, I ask it very, very specific questions, and then I run everything through GPT zero so it doesn't think anything was written by by the LLM, right?
• 11:39
  SPEAKER_01: Because I'm very, very careful about that.
• 11:41
  SPEAKER_01: I'm a big believer in in doing it from my mind and my creativity, but using it as a tool.
• 11:47
  SPEAKER_01: And then if I do quote from from uh an LLM, I will actually annotate it.
• 11:52
  SPEAKER_01: I will say this section is verbatim from the LLM.
• 11:55
  SPEAKER_01: And I think that's what's important because a lot of times people, you know, they're they're passing this stuff off as creative content or new content or you know, thought leadership, and it and it really is not, it's just a regurgitation of you know, the stuff from our stochastic parrot, right?
• 12:11
  SPEAKER_01: Which is what the LLMs are.
• 12:13
  SPEAKER_03: Um, so so let's that was definitely an SAT word right there.
• 12:20
  SPEAKER_01: So tell me a little bit about, you know, we we talked about Google's ability, and of course, the large player's ability, and Google has what, all uh 15 exabytes, I think was the was the figure you quoted.
• 12:32
  SPEAKER_01: First of all, what does that look like?
• 12:34
  SPEAKER_01: How can you get your mind around what 15 exabytes is?
• 12:38
  SPEAKER_01: And then how is that going to be leveraged?
• 12:41
  SPEAKER_01: And then uh this is a three-part question.
• 12:44
  SPEAKER_01: And then what do organizations need to do uh in terms of SEO when you have all of these changes in in algorithms in the marketplace?
• 12:55
  SPEAKER_03: Yeah, so um, you know, several of our businesses, we own three holding companies uh that are involved in uh search engine optimization, largely uh across uh the link building agency kind of model, which is really about finding brands and then connecting them with other brands and then looking at ways in which they can um benefit each other uh in a non-competitive way.
• 13:20
  SPEAKER_03: And so, you know, I think even just recently, you know, all of the uproar uh regarding open AI and um uh Mr.
• 13:29
  SPEAKER_03: Altman and that whole situation, it's kind of showing that the um, so I'm gonna start first with like Fang plus Salesforce, right?
• 13:40
  SPEAKER_03: So all of these ecosystems, uh whether it's Google, whether it's Microsoft, you know, Excel, Word, you know, Office, whether it's Facebook, um, and and whether it's you know Amazon.
• 13:54
  SPEAKER_03: And then I put Salesforce in there as well because Salesforce is you know really deployed across the board as as a key uh uh platform within almost every organization.
• 14:07
  SPEAKER_03: And uh they're coming out with Einstein in terms of their marketing at uh Pardot, and they have Einstein coming out uh that we're going to be we're gonna be leveraging.
• 14:17
  SPEAKER_03: But um, and so if you look at these ecosystems, Chat GPT and these other versions of that are a threat to their ecosystem because they have millions of users that are you know tied into their systems.
• 14:34
  SPEAKER_03: And if you start to allow a foreign kind of entity into that ecosystem and progressively it makes those tools less and less valuable, and you start losing people out of those ecosystems, and that that is what really is driving, I believe, most of the decision making right now is almost like a land grab in terms of securing and and buttoning down your ecosystem and then defining AI within the context of your ecosystem and creating your own models, even with what Elon has done with you know X, and to ensure that you're keeping the activity within your ecosystem as opposed to losing them to you know open AI and and who knows where where those those customers go at that point.
• 15:19
  SPEAKER_03: And and so search generative experience has been ongoing.
• 15:23
  SPEAKER_03: So it used to be that Google would provide like updates saying, hey, we're doing a panda update, or we did this, and and really search generative experience is not an update, it's a it's a it's a it's a C change and it's it's continuous, it's an infinite game as opposed to uh a finite game in terms of I believe the way that Google is viewing um the threat of open AI chat GPT.
• 15:48
  SPEAKER_03: And so what they have is 15 exobytes of data, uh it's estimated.
• 15:54
  SPEAKER_03: Nobody knows for sure, but I I did my research and and and it's probably even a little bit larger in my opinion.
• 16:01
  SPEAKER_03: But if you think about that in the context of if you were to have recorded every human conversation since the dawn of time, every human conversation since the dawn of time, it would account for about five exobytes of data.
• 16:17
  SPEAKER_03: So you you can imagine that Google has immense information, not only and it's anonymized on a personal level, but at the same point, they're getting all of these signals from all these personas.
• 16:29
  SPEAKER_03: They see all the search activity, they have you know Gmail, they have all these different ecosystem, and and that is incredible computational uh technology.
• 16:40
  SPEAKER_03: And now you can imagine that they're layering on and leveraging that computational infrastructure to be able to uh really define um uh the the the use of AI to deliver you know truly personalized search experiences.
• 16:58
  SPEAKER_03: And and in and an illustration of that is if you're you know, if you and I uh are you know in the tech space, this, that, and the other, and we're in two locations, we may relatively the same age, etc.
• 17:13
  SPEAKER_03: Like we may get uh you know, we may get a relatively uh You're in your late 40s like me, right?
• 17:20
  SPEAKER_03: Yeah, yeah, yeah, yeah.
• 17:22
  SPEAKER_03: Relatively the same type of search result when we're going out and doing searching.
• 17:27
  SPEAKER_03: And um, and that's because we are defined within a persona, right?
• 17:32
  SPEAKER_03: Like um, you know, very handsome, uh uh bright guys, you know, whatever.
• 17:38
  SPEAKER_03: I'm kidding.
• 17:39
  SPEAKER_03: Um, but and uh and and so basically we would get that search, but but imagine now that they can actually go down to the level of of you as a persona, you as an individual, and right signals.
• 17:53
  SPEAKER_03: Yeah.
• 17:53
  SPEAKER_03: And then also, if you'll note, like if you're using Google Labs, like it, they're starting to finish your sentence, not not like uh, not like it used to be, where you'd start searching for pizza near me, and it would be like you type in pizza and then it would say near me.
• 18:08
  SPEAKER_03: Now your search phrase is like uh, you know, what's going on at OpenAI?
• 18:13
  SPEAKER_03: It it may with and then it may append that with like five or six more things.
• 18:18
  SPEAKER_03: And I'm finding, or in other words, um, or complete that sentence, and based upon all the signals that they're getting, that that's what people really want to find out about what's going on at OpenAI.
• 18:30
  SPEAKER_01: So, what does that mean for traditional search engine optimization, right?
• 18:35
  SPEAKER_01: Because it's not just a matter of keyword density and it's not just a matter of linkages.
• 18:40
  SPEAKER_01: Uh, it has to do with really um nuanced details, uh, the signals, the digital body language of uh users.
• 18:48
  SPEAKER_01: Now, here's something I refuse to do, like no matter how many times I'm on my iPhone, and for some reason it's it, you know, when I ask a question, it comes up and it says, gives me Google Maps, and then it's like sign in and use the app.
• 19:02
  SPEAKER_01: And I'm like, absolutely refuse, absolutely refuse.
• 19:05
  SPEAKER_01: And it's really hard because I'm always having to take extra steps to not sign in because I just, you know, I just feel it's intrusive, you know.
• 19:13
  SPEAKER_01: Um, so I'm trying to give it fewer signals and trying to, you know, but Apple is doing the same.
• 19:18
  SPEAKER_01: But what does this really mean for you know organizations that are trying to maintain their, you know, their uh inbound uh, you know, traffic and trying to optimize their content?
• 19:29
  SPEAKER_01: So how do they need to be thinking of this or reacting to this?
• 19:33
  SPEAKER_03: Yeah, so um look, it's it's estimated that uh there's going to be a decline, or people should plan for a decline of about 25% of their organic traffic.
• 19:46
  SPEAKER_03: And the key is is that like, you know, we have we've served over 270,000 customers in the last like eight or nine years.
• 19:54
  SPEAKER_03: We have you know 270,000 users.
• 19:57
  SPEAKER_03: And so, you know, what we've known when we've we've done a lot of work with them in terms of identifying and and what it comes down to is really setting yourself up for success by making sure that that 25% decline comes from people that weren't ever going to do business with you in the first place.
• 20:15
  SPEAKER_01: Right.
• 20:16
  SPEAKER_03: And so the way that you do that is really defining who your right fit customer is, and then who, and then identifying what that avatar or persona is at a at a very, you know, a very deep human level, and then writing your content and your positioning to those right fit customers so that Google says this is being written to this type of avatar, and it's providing value to that avatar.
• 20:46
  SPEAKER_03: And so therefore, we're going to place it higher in the search rankings for that, for that, uh, for that individual.
• 20:53
  SPEAKER_03: And so, um, and then don't even get me on like user experience on on the back end.
• 20:59
  SPEAKER_03: Like I, you know, you you're gonna have to create and continue to focus on much more uh friction-free ways to uh connect with your with your prospective customers and meet them in the moment, as opposed to like expecting somebody to go do a calendar, you know, and match it up with their calendar and book a meeting and and and and go through that whole process from a B2B perspective.
• 21:25
  SPEAKER_03: So I think that it's it's going to be uh it's gonna be a long needed uh adjustment to how marketers like, I mean, how many times do you land on a website before you've even gotten the value out of why you got there, they're asking you to sign up for their email list.
• 21:40
  SPEAKER_01: All the time.
• 21:41
  SPEAKER_01: It's like a sentence.
• 21:44
  SPEAKER_01: Yeah, and I bounce and I bounce.
• 21:46
  SPEAKER_01: I bounce.
• 21:47
  SPEAKER_01: You know, I close it, and if they do something else, I bounce, right?
• 21:51
  SPEAKER_01: Or oh, let's let's take all your cookies.
• 21:53
  SPEAKER_01: I love the ones that give you the option.
• 21:55
  SPEAKER_01: Do you want to accept all the cookies, or do you want to look at our cookies?
• 22:00
  SPEAKER_01: Policy.
• 22:00
  SPEAKER_01: Yeah, like I don't want to accept your cookies, and you're not giving me an opportunity to tell you whether I can or not.
• 22:07
  SPEAKER_01: So anyway, but um, but that's that is that's very clear that the user experience is being impacted.
• 22:13
  SPEAKER_01: And but it sounds like we also need to think about as marketers, marketers need to think about multiple dimensions of the uh content and and the user in order to to be more more uh nuanced and more targeted for what they're doing.
• 22:29
  SPEAKER_01: So it's a much more complex um uh set of optimization parameters, it sounds like.
• 22:35
  SPEAKER_03: Yeah, it reminds me of Star Trek, the the chess, the four-dimensional chess board, you know.
• 22:40
  SPEAKER_03: Yes, it's no longer, and and you coined a phrase in our in our prep call a few weeks ago.
• 22:45
  SPEAKER_03: It was like I forget what it was.
• 22:47
  SPEAKER_03: I I told you that I was going to uh uh ask if I could use it.
• 22:51
  SPEAKER_03: I'll get it.
• 22:51
  SPEAKER_03: Yeah, but now we can't remember it.
• 22:53
  SPEAKER_03: I think it was a multi-dimensional, it's a multidimensional or something.
• 22:57
  SPEAKER_01: And yeah, there's something about multidimensional.
• 22:59
  SPEAKER_01: Maybe Liam can text it to me in the uh uh in the chat here if he remembers what it was.
• 23:04
  SPEAKER_01: But I I did come across that.
• 23:05
  SPEAKER_01: Now I can't remember the context, but it it it it does have multiple multiple dimensions, and uh it is an interesting problem space.
• 23:13
  SPEAKER_01: Now let's shift gears a little bit and talk about you know, when you start thinking about um uh you know AI-powered uh agents and you know how we're trying to offload some of those you know, rudimentary or basic or repetitive uh tasks, password changes.
• 23:31
  SPEAKER_01: Nobody wants to call anybody for a password change.
• 23:33
  SPEAKER_01: You want a password change, right?
• 23:35
  SPEAKER_01: But but but you know, there's a lot of tasks that are being automated that normally would be say entry-level positions for uh for uh you know graduates or even people that didn't graduate from from college, right?
• 23:48
  SPEAKER_01: What are the what what does this mean for career paths?
• 23:51
  SPEAKER_01: What does this mean for uh how people need to be prepared for uh the the job market both now and in the future?
• 23:59
  SPEAKER_03: Yeah, I mean it's uh I I certainly uh uh I certainly don't have uh I don't have a way of of looking into the future, but I can tell you what what I'm what we're doing um to help our in employees and to uh help prepare the next generation.
• 24:18
  SPEAKER_03: I mean, I have a 20-year-old daughter and a 17-year-old son, so I'm I'm also keenly, you know, looking at it from a context of of where where where where should they go in their professional career?
• 24:30
  SPEAKER_03: And um I think I think number one, you know, there's a lot of talk about prompt engineering, but I mean, if you look at it now, you can you can train you can train chat GPT to to not to try not to think of itself as chat GPT, right, and then to create the prompts for what you want to do.
• 24:53
  SPEAKER_03: So it's almost like it's become the prompt engineer.
• 24:57
  SPEAKER_01: And so you're giving it a persona from which to think about answering those questions or think or to frame those questions or get in the right vector space, right?
• 25:09
  SPEAKER_01: In order to, you're saying you are this, and that automatically changes where it's going to be looking in the large language model for certain things.
• 25:18
  SPEAKER_01: You're getting closer to what as an expert in you know marketing automation, answer the following questions, you know, as a critical tech columnist or something along those lines.
• 25:28
  SPEAKER_01: I I did that with it to say find the flaws in my argument, right?
• 25:32
  SPEAKER_01: And and that was very valuable.
• 25:34
  SPEAKER_03: And I think that so if if that is if if we're if that's where it's going, then who is going to be the human that is defining what that persona is that the that and and furthering the and and and providing the the intellectual input to be able to to to to craft that in a way that that achieves the the task at a high level?
• 26:01
  SPEAKER_03: And then who's gonna be the person that is going to be managing uh the the the boss you know the the autonomous agent and and ensuring that it's not it it doesn't start out really good and then as you mentioned before all of a sudden start going off the rails?
• 26:20
  SPEAKER_03: And if you have automatic, if you have automations with this stuff, no, like I mean I I think like you know I think about like marketing, like like email marketing or others, like you know, sequences have been like the thing, right?
• 26:36
  SPEAKER_03: But have you ever like like when you run when you have these sequences running, like once you start them, like you you can't turn them off.
• 26:45
  SPEAKER_03: Like we basically all you can do is just turn them off and stop them if somebody responds, stop right.
• 26:52
  SPEAKER_03: But but a lot of times those things you know they get bigger and bigger and bigger, and then they and and they and they start failing it almost in in in in unpredictable ways.
• 27:02
  SPEAKER_03: You can have one of those spin out, and it can knock out your entire marketing.
• 27:08
  SPEAKER_03: Uh, and and I use that example.
• 27:09
  SPEAKER_03: We used to have a lot of automations running in Salesforce, and we had you know 120 salespeople, and all of a sudden, some cron job that we had written, you know, a year and a half ago to keep everything working started to spin out of control with an update, right?
• 27:26
  SPEAKER_03: And it knocked out our ability to do uh selling for like three or four days.
• 27:31
  SPEAKER_03: And so I mean, you can imagine when you're I mean, like it can take out your whole business.
• 27:36
  SPEAKER_03: So that's where I think that the the the responsible agent, the responsible human agent managing the autonomous agent and ensuring that it and really just managing it as if it's a not as if it's a person, but managing it, making sure that it's ethical and and moral in what it's producing, make sure it's not lying.
• 27:55
  SPEAKER_03: I mean, you know, AI likes to lie, like like it it it's so it'll just tell you what it thinks you want to hear.
• 28:02
  SPEAKER_03: And so right, those types of things I think are the are the the real scary part.
• 28:07
  SPEAKER_03: Um not the scary part, but I think that that that's the risk, right?
• 28:10
  SPEAKER_03: Is is that these things start running out of control and then nobody's managing them.
• 28:14
  SPEAKER_01: Right.
• 28:15
  SPEAKER_01: So instead of a helper bot, I have a helper human.
• 28:18
  SPEAKER_01: His name is Ian, uh Liam, sorry, sorry, Liam, and uh he did say it was multidimensional targeting, is what we had talked about last time.
• 28:26
  SPEAKER_01: So when you think about all of those pieces, we have to think about all the dimensions that describe that individual, but then where they are in their journey.
• 28:34
  SPEAKER_01: And that's actually uh sometime you and I have to go through my my orchestration model, which actually talks exactly about that um in terms of uh taking that digital body language or those signals from from customers or employees, and then serves up the content.
• 28:50
  SPEAKER_01: But again, we have to do that on our content for public consumption and for uh organic search.
• 28:56
  SPEAKER_01: So, yes, so it's the multidimensional targeting that we had talked about.
• 29:00
  SPEAKER_01: So um, so when you think about um, you know, trying to uh have that balance between privacy concerns and leveraging user data, um, you know, how do you see that being balanced uh in in its ecosystem, you know, versus staying competitive with Microsoft and Apple, right?
• 29:25
  SPEAKER_01: Because it has all of this advantage.
• 29:27
  SPEAKER_01: But you know, sometimes, as we say, you know, when you when you get when it when it gets a little too intimate, right, it gets a little creepy and uh it goes through that uncanny valley.
• 29:37
  SPEAKER_01: I know that is used to be referring to animations and computer generated uh uh imagery, uh, where it wasn't quite human, but it wasn't close enough to being human to really just be creepy, right?
• 29:50
  SPEAKER_01: And the same thing is happening with with privacy, and I think organizations need to think about that as well, right?
• 29:55
  SPEAKER_01: We we know the the famous case of target, you know, sending um, you know, expectant mother uh marketing materials to uh a daughter within a family, and they're like, Why are you sending this to my daughter?
• 30:08
  SPEAKER_01: And it's like because they knew she was pregnant before the parents knew she was pregnant, right?
• 30:12
  SPEAKER_01: Because other signals.
• 30:14
  SPEAKER_01: So, you know, that's going to be a perennial problem and uh obviously going to get more challenging.
• 30:19
  SPEAKER_01: And of course, the regulation is trying to keep up with it.
• 30:22
  SPEAKER_01: But do you have any thoughts on you know balance making uh you know balancing and making those those trade-offs uh decisions?
• 30:29
  SPEAKER_03: You know, I I look at the you know, like the Altman situation, and and I think that the only thing that the board had released was that there was a lack of candor, uh candor.
• 30:39
  SPEAKER_01: Yeah, I haven't seen anything else on that.
• 30:41
  SPEAKER_03: So like I I I mean that's that by the way scares the heck out of me, but but besides that, but you know, it it is you know, corporations are you know meant to make money, at least in you know capitalist societies.
• 30:59
  SPEAKER_03: And you know, it we're really entrusting uh a lot of these, a lot of the the the executive management um to be doing the right thing as it relates to their ecosystems.
• 31:11
  SPEAKER_03: And I didn't mention Apple earlier, but obviously Apple's uh a key player in that as well, but on the thing.
• 31:17
  SPEAKER_03: But um, and I and I I I you know I I have very little faith, first of all, in you know, an act of Congress happening, or jokingly, I mean, or that they're going to be able to stay on top of this to the extent that they they need to be on top of it.
• 31:37
  SPEAKER_03: Um I think that I think Great Britain uh has actually been the UK has actually been, you know, they held their AI conference.
• 31:47
  SPEAKER_03: Um, and and I I I've gotten a lot of sense that the EU and Great Britain have obviously been much further along in terms of trying to protect uh privacy and security of the of the users.
• 31:59
  SPEAKER_03: Um and but at the end of the day, like you know, us as users, you know that every time you're putting something into chat GPT, you you're you're making it available for the broader audience.
• 32:13
  SPEAKER_03: Um we have a we've had a direct pipe to open AI since since the very beginning.
• 32:18
  SPEAKER_03: Yeah and that's supposed to provide us some layers of protection for our company.
• 32:23
  SPEAKER_03: But I'll give you an example.
• 32:25
  SPEAKER_03: So when Chat GPT, you know, Turbo came out and nearly broke the internet, um, you know, it was wild.
• 32:32
  SPEAKER_03: But uh I was like a little freaked out uh that morning.
• 32:35
  SPEAKER_03: I was I was worried that maybe singularity had happened.
• 32:38
  SPEAKER_03: Um and and then everybody was looking at me around the table like what's singularity, and I was like, you don't want to know.
• 32:44
  SPEAKER_03: Um, but the the whole point was it's like you know um we we have to be responsible human beings, like responsible users of technology.
• 32:56
  SPEAKER_03: Like if we're continually putting ourselves into these models and uh and and and and and giving that away, and particularly even on a business setting, I mean when GPT before turbo came out or whatever they're calling it, I I was like, oh, I mean, and I knew that there was wrapper products.
• 33:16
  SPEAKER_03: There's like what 5,900 wrapper products that have come out in the last like nine months that were just basically prompts, pre-built prompts.
• 33:24
  SPEAKER_03: But you know, I it was like it had the little thing where I could upload a spreadsheet, right?
• 33:29
  SPEAKER_03: And I was like, I was tempted to basically upload all my customer data and then just to start anonymized, but then just to go through and start asking it questions who's my right fit customer, who's more likely to churn, who's this, this, this, and this.
• 33:43
  SPEAKER_03: And then I was like, Oh, yeah, I I'm gonna be uploading, and I would never upload customer data, but I'm just saying, like, like, like purchase date, like that, you know, it's like and I was like, absolutely not.
• 33:55
  SPEAKER_03: I mean, we don't have any personal information, so sure.
• 33:58
  SPEAKER_03: Um, but I was just like, wow, maybe I could quickly get some answers.
• 34:02
  SPEAKER_03: And then I was like, no way, you know, no way.
• 34:06
  SPEAKER_01: Well, it does, it does beg the question of you know, how to protect intellectual property and competitive advantage and get the best out of these large language models, because the LLM, the you know, asking the LLM the question is one thing, and it'll answer the question based on its understanding of language and concepts and terminology.
• 34:25
  SPEAKER_01: But that's when we can be prone to hallucinations.
• 34:29
  SPEAKER_01: But if we instead, and I think we've probably talked a little bit about retrieval augmented generation, did we cover that a little bit?
• 34:35
  SPEAKER_01: Yeah.
• 34:35
  SPEAKER_01: So the idea of you know using the LLM to interpret the question, uh, to then use that for to to uh run against uh run a query against your content, then your content could either be in a knowledge base and you're using uh semantic search or or metadata search, or you can ingest that content into a vector database, a vector space, and use the metadata for enriching those embeddings.
• 35:02
  SPEAKER_01: And those become additional signals for the LLM to answer the questions, and then using the result, uh processing the result with the LLM so that it can make it more conversational.
• 35:14
  SPEAKER_01: And if you turn the temperature down, you know, the creativity parameter, the temperature from you know, one to zero, right?
• 35:22
  SPEAKER_01: One being more creative, zero being less creative, and you say, you know, if you don't have the answer from this data source, say I don't know, then that's really protecting a lot more.
• 35:32
  SPEAKER_01: But it's also allowing organizations to use that um that uh competitive differentiation, right?
• 35:38
  SPEAKER_01: Standardization will give you efficiency, differentiation gives us competitive advantage.
• 35:43
  SPEAKER_01: But as you say, you know, we have to be very careful about that.
• 35:46
  SPEAKER_01: Even with the API level stuff, that's what they say, but I don't know if every organization is going to trust that, right?
• 35:53
  SPEAKER_01: So you have to use localized language models for that.
• 35:57
  SPEAKER_01: So have you seen um organizations kind of looking at that or understanding it, or is it still too early in the marketplace for really people to understand the nature of the problem and what the solution would look like?
• 36:09
  SPEAKER_01: I I find that that it's a it's you know, when you start going back to retrieval and you start saying metadata or ontology or knowledge graph or things that they think the the uh technology has obviated, right?
• 36:25
  SPEAKER_01: Has has eliminated.
• 36:27
  SPEAKER_01: Um people kind of get a little, you know, frowny face, right?
• 36:30
  SPEAKER_01: Because they don't want to do that work because they thought it was a silver bullet.
• 36:33
  SPEAKER_01: Do you do you have any opinions on that or any thoughts on that?
• 36:36
  SPEAKER_01: Have you seen it much, or is it still too early uh in the marketplace from what you're witnessing?
• 36:42
  SPEAKER_03: I it's almost like, you know, it's almost like the um I I I'm trying to think of a way to describe it, but it's like we're you know, we're we're I think there is a a layer of of management, you know, executive level that that is that is uh uh maybe, maybe I always say, you know, it's good to have a healthy sense of paranoia, you know, and and is and is acting on that.
• 37:11
  SPEAKER_03: But you know, the I don't know, like I've sat in a lot of different uh executive meetings with you know the CFO and and et cetera.
• 37:20
  SPEAKER_03: And it's like, you know, anytime IT talks or anytime that marketing talks, typically everybody just thinks that most of what they're saying is is nonsense.
• 37:29
  SPEAKER_03: Um and they're just trying to get a bigger budget, right?
• 37:32
  SPEAKER_03: And and oftentimes that's been true over the years.
• 37:36
  SPEAKER_03: But as you start entering, like, you know, I employed at a previous company two PhDs, you know, and and it wasn't the 24 million annual reoccurring revenue marketing automation SAS, but it wasn't until we were able to to leverage the power of those PhDs to um to be able to unlock and and to move us forward from a from a data perspective, single point of truth, et cetera.
• 38:00
  SPEAKER_03: So even the the the words and phrases that you used, I I can see that I I don't know that there's leaders within, I haven't seen a lot of leaders within companies that even have an appetite for trying to trying to understand it.
• 38:18
  SPEAKER_03: I think they're just trying to to maybe use it, use AI in a way that provides that efficiency um versus the different differentiation.
• 38:29
  SPEAKER_03: Yeah, that's that's what I see.
• 38:30
  SPEAKER_03: It's it's all about efficiency as and it's not about leveraging it for a differentiated yeah.
• 38:38
  SPEAKER_01: Yeah, yeah.
• 38:39
  SPEAKER_01: And well, there are the analyst firms such as Orster that say retrieval augmented generation is going to have the single largest impact of any AI technology on the enterprise.
• 38:51
  SPEAKER_01: Now, what the timeline for that is, I don't know, but it's making all of that information more readily accessible and reducing the friction of info or information flows across the organization.
• 39:01
  SPEAKER_01: I did a briefing for about uh 20 or so bank CEOs uh not too long ago, where I explained these concepts and they were actually very receptive to them and understood them.
• 39:12
  SPEAKER_03: Yeah, I think you're you're I mean, I was just gonna say if if that's what you're if you're able to be an evangelist and to get in it, just needs to have more people like you getting in front of more audiences and and you're trying to do it through your podcast, you're you know what I mean, but like that's what it's all about is is is people with your intellectual capacity and and experience, but then also your ability to frame it in a way that makes sense for uh uh uh Steve, the the the CFO, uh that you know, uh to his kids' soccer games on the weekend and has a full-time job.
• 39:49
  SPEAKER_01: Uh other people if you could only get through all the noise and nonsense and the inundation of all of the uh uh you know misconceptions and you know the vendors that are all jumping onto this and and chasing it down.
• 40:06
  SPEAKER_01: I think I might have told you about the vendor who uh said uh you know, they were doing a customer service application.
• 40:12
  SPEAKER_01: They said they said, So tell me about the role of you know ontologies, metadata, information architecture, taxonomies, whatever term I use, because oh, you don't need any of that.
• 40:21
  SPEAKER_01: They said, Really?
• 40:21
  SPEAKER_01: What do you do with the data?
• 40:23
  SPEAKER_01: Well, you do have to do some data labeling.
• 40:25
  SPEAKER_01: It's like, duh, data labeling is information architecture, that is metadata, that is that uses a taxonomy and a reference architecture.
• 40:32
  SPEAKER_01: So I think a lot of folks are not necessarily don't have the history to understand the core principles here and to make those connections and to help organizations understand what they actually have to do.
• 40:43
  SPEAKER_01: But then again, maybe they do and they just don't want to, you know, they just want to keep people in dark.
• 40:47
  SPEAKER_01: You know, like if you explain and it sounds simple, it's like, oh, it can't be any magic there.
• 40:52
  unknown: Yeah.
• 40:53
  SPEAKER_03: Right.
• 40:54
  SPEAKER_03: Right.
• 40:55
  SPEAKER_03: I I think you're doing good work, and I think that that really is the that's what needs to happen, is it's just it that you know that it just getting in front of more people, cutting through the noise, and and and explaining things in a way that uh makes sure that they're they're doing the appropriate steps as they engage AI and that there are is there is work that needs to be done in order to leverage it effectively.
• 41:23
  SPEAKER_01: Feel free to bring me in to any of your customers to help with that.
• 41:29
  SPEAKER_01: I'll do that uh a comp on a complimentary basis for you.
• 41:33
  SPEAKER_01: So let's go, let's shift uh gears a little bit.
• 41:36
  SPEAKER_01: Um I think you mentioned that you're a tennis fan and a bit of a biohacker.
• 41:41
  SPEAKER_01: Tell me a little bit more about that.
• 41:42
  SPEAKER_01: Tell me uh how you're you're you're uh dealing with fitness and if there's a role of technology in there and give me a sense of uh that personal side.
• 41:52
  SPEAKER_03: Yeah, I I think uh I think actually some of the the larger uh investments uh that are coming from AI from the private equity and and venture space have been uh quite a bit of a shift into healthcare related.
• 42:05
  SPEAKER_03: And you know, it's uh it's it's a frightening um you know, it it's it's frightening in terms of the you know, the amount of uh of what is a 25% of our GDP is is healthcare.
• 42:17
  SPEAKER_03: Uh 42 percent of the uh of discretion of of the uh copay uh for uh hospital visits and healthcare visits goes unpaid.
• 42:29
  SPEAKER_03: Um and so yeah, like they just don't collect it because it it it you know, and I don't I I I so you know my my kind of goal I pay all my co-pays, maybe I should stop doing that.
• 42:40
  SPEAKER_03: Well, yeah, and and it's just because people don't, anyways, I won't get into that, but I'm just saying that like healthcare obviously is is enormously uh uh uh uh bloated with both regulation overlays and then just the fact that it's it's an incredibly personal and and and and and deeply personal uh interaction.
• 42:58
  SPEAKER_03: And so um a few years ago, uh well, probably about 10, uh, you know, I uh uh I was on a nonprofit uh uh with a good friend of mine, and and and he's a physician.
• 43:10
  SPEAKER_03: And we really got into um, you know, this is uh we really got into to to looking at uh the the healthcare clinic helped people 200% below the poverty line.
• 43:21
  SPEAKER_03: And what we were seeing is that you know, diabetes and and all the related, like you, you know, you get diabetes, you hype have high blood pressure, then you have diabetes, then you have to take, you know, then you have cholesterol, then you have sexual dysfunction, then you have this and this and this and this and this.
• 43:36
  SPEAKER_03: So now you're you're taking you know 10 different medications that all relate to the fact that you're just not most probably living a healthy life.
• 43:45
  SPEAKER_03: And I'm not saying that means you're you're weight or whatever, but you can be of a certain size and still be super healthy.
• 43:52
  SPEAKER_03: And so, you know, we started uh doing some really interesting stuff in terms of like even a a a a a uh uh glycolic uh uh uh glycemic uh meters, uh blood sugar meters, and then testing different foods and watching how our blood sugar was spiking and then designing almost like personalized uh I hate the word diet, but food plans, food and nutrition plans that were based upon optimizing our energy levels over the course of the day.
• 44:24
  SPEAKER_02: Yeah.
• 44:25
  SPEAKER_03: Um and so that that's just kind of set me and then I I've written on like you said we, that's you personally kind of in your family during yeah yeah and I mean like you did and then uh you know my my kids it we they will anyways uh I was going to tell a joke about the day that I had to go buy some ingredients for uh my kids uh uh elementary school party and the teacher assigned me uh Oreos and and like Sunny D and I I was so so I try to eat less than five grams of sugar a day and uh sugar is like is is such a you know force and and so I was literally like walk like putting the thing on the on the counter and I was checking out and I was almost apologizing to the cashier saying it's you know it's not for me I promise and and I told her friends yeah I said I can't do this again like it just because I feel so bad that I'm dumping this on these children.
• 45:20
  SPEAKER_03: I'm only seeing the problem.
• 45:22
  SPEAKER_03: So anyways that's that's kind of the biohacking and then I yeah I tried to keep myself at a 48 to 47 a metabolic age and um and and and you know keep my keep everything kind of in line.
• 45:34
  SPEAKER_03: How how is that what are the biomarkers for that by that uh biological age like you know uh BMI uh it's uh muscle mass it's resting heart rate resting heart rate water water weight um you know bone density bone density um it's like a myriad of those different things and then it yeah it's not look it's it is it am I 47?
• 45:58
  SPEAKER_03: No but you know it's at least it provides a baseline and a metric from which I I was 54 and I'm only 52 when I started at one point I had some three and so I was my goal was to get to the other direction.
• 46:11
  SPEAKER_03: And so at least it tells you that you're making progress in the right direction.
• 46:15
  SPEAKER_01: As I say you're in your late 40s uh your very late 40s you're 4014.
• 46:20
  SPEAKER_01: 4014 okay it's a math problem in base in base 12 you're even younger that that joke doesn't go over well in no I conferences years you exceeded my uh my it's a math problem an intellectual humor level so um so of course and uh that's great because you know your your physical health is so important in terms of running a business because you need lots of energy for that and you need to keep a positive output.
• 46:49
  SPEAKER_03: Um so here's a here's a quick um uh question kind of an introspective question if you were to kind of go back and think of yourself as a graduate uh of school what what kinds of advice might you give yourself uh at this point uh you know for with with this retrospective understanding what advice would you give yourself getting out of school um I mean I would probably frame it more unfortunately it's it's probably hard for me to do that given how much the world has changed you know like the rise of the internet came after school like I mean so I I would and I I do a lot of thinking on this in terms of you know 40% of middle management in the United States are millennials um and my my son and daughter are two different or of the Gen Z and et cetera and you know there's there's I I would just say like so if I were today in today's world um you know there's a lot of cynicism uh about just about everything and you know unfortunately uh you know when people tell me that uh oh oh this AI thing is a fad I don't I don't you know I don't it's whatever I I tell them I say do you understand that most of your friends and your political views have been shaped by AI like on social media it has been running and and determining your point of view hijacked us hijacked us it's been hijacking and and so you know and so you know my kids and even the the the members of our our team it's certain members of the team and and and you know there's a lot of cynicism towards um uh towards technology you know a lot of these kids have like you back to what you said have just been targeted with the most ridiculous ads on YouTube and they you know or they've or they've had a synchronicity to where all of a sudden the ad is like right on point for something that and and I just find that there's a lot of cynicism about technology and and I've always held the view that it it can it it can be such a force for good and and in in kind of the basis consumerism technology AI you know if you get above that there's really there's some wondrous things that can take place for humankind and they can be a part of that.
• 49:16
  SPEAKER_03: That's so I would say don't be a cynic lean in our better angels of technology and be a part of the solution don't sit back and say oh this all is terrible and and it's too big of a monster we'll go tame the monster you know who else is gonna solve it you know right uh today's graduates and tomorrow's graduates and the people who are going to you know inherit the inherit the the earth inherit you know whatever is you know whatever we end up leaving uh in based on the decisions we make in the next uh couple of decades right my favorite my favorite quote is uh it's a chinese proverb that says every hundred years all new people so you got to think under every hundred years what all new people all new people yeah everybody yep you're right we turn over everything um well mark it's been a real pleasure so uh where can people find you we'll put this in the show notes as well but um uh what's the best way to reach you yeah so uh uh you know LinkedIn's a great way to reach me um I'm uh I I was early to LinkedIn so it's uh you know Mark Pickren uh is the uh I'm listed M-A-R-C P-I-C-K-R-E-N and then um you know uh we're we're uh you know WWW I I'm an old guy I actually do that these days but uh next netmedia.com uh is where you can and there's actually a connection point on uh uh I think on our executive team to where you can actually connect with me because I'm I'm just really about like just I I'll talk to anybody I'm all about trying to be helpful to folks um and uh that's you know that's kind of where I I sit right now so super appreciative of you and uh an honor to be on the show.
• 51:06
  SPEAKER_01: Thank you so much for being with us I really appreciate it.
• 51:10
  SPEAKER_01: All right it's been great and so um uh again thank you it's uh it's been tremendous really enjoyed the conversation so this has been another episode I have to say exciting episode because I really enjoy it uh of the early AI podcast and uh we will see you all next time thank you very much all right bye for now thank you for joining us on another deep dive into AI innovation tune in next time when we introduce another industry expert and discuss how to maximize AI in your world the early AI podcast is sponsored by early information science and CMS Wire.
• 51:48
  SPEAKER_00: To learn more visit early.com that is e-ar l-e y dot com thanks for tuning in