Transcript 0:00 Welcome to the debate, We the People in Tech or Not? The debate, to be exact. 0:05 Um, I have to mention right at the top that this is an AI-generated podcast based on a post from the We the People in Tech or Not newsletter by Alex Rosa. 0:13 And the goal of this show is to help listeners learn, you know, new ways to explain technical topics ranging from a beginner non-tech understanding up to a fully expert level. Exactly. 0:22 And today we are unpacking a really core issue from that newsletter. Basically, AI search tools are fundamentally changing how people look for information. They do not give you Tim Blue links to browse anymore. 0:33 They give you three to five definitive names and, well, if you aren't one of them, you are essentially invisible. It is a pretty wild shift in how search works. It really is. 0:44 So I will be defending the position that taking active control of your AI visibility is just an essential, highly effective step for any professional or business today. And I will be playing the respectful skeptic here. 0:56 I wanna represent the smart, you know, busy non-tech manager listening on their commute. I am really questioning if this level of technical effort is truly practical or necessary for the average person. 1:07 That is totally fair. Yeah, I mean, given the costs, the risks, and just the limits of what we can actually control on the internet, there is a real debate about the return on investment here. 1:17 Well, let us jump right into that return on investment because the material has a really striking example. It talks about Marina Mogilko, known online as Silicon Valley Girl. 1:26 She realized AI tools were just completely ignoring her podcast. Which is incredibly frustrating, I am sure. Oh, absolutely. Yeah. So instead of just accepting it, her team took these deliberate plain language steps. 1:38 They gave every single podcast episode its own plain text page so the AI could actually ingest the transcripts easily. Okay, that makes sense. Right. Just simple text. Right. 1:47 And they found this outdated database calling her a vlogger from ten years ago and updated that public fact across eleven different languages. Then they explicitly submitted those pages to Google. 1:57 And what was the actual payoff for that kind of effort? It was huge. Google found ninety-one pages it had never indexed. She gained, uh, about seven hundred new clicks a month. 2:07 But the best part, tracking forty-eight specific questions across eleven podcast brands, she went from showing up in only seven of those answers to becoming the fifth most visible brand overall. Wow. Fifth most visible? 2:19 Yeah. By organizing her data clearly, she completely shifted the algorithm's weighting. That's quite a jump. But, you know, we really have to look at the reality of the effort required here. 2:29 Let us look at Alex Rosa, the author of the post. Sure. Let's look at his experience with this. He is a veteran tech professional, right? 2:35 Over thirty years of experience, a senior cloud solution architect, and yet the post says he waited five weeks to act after watching Marina's video. Well, yeah, he wanted a, a good architecture first. Exactly. 2:49 He felt he needed a good architecture. 2:51 If a thirty-year tech veteran needs a five-week incubation period just to clear the mental and technical hurdles for a simple website, I mean, that is a massive barrier to entry for, say, a local bakery owner. 3:03 I hear you. It definitely sounds daunting. And it gets worse when you look at another statistic from the post. Eighty-two percent of what an AI cites is earned media. Right. The external ecosystem. 3:14 Yeah, that is the stuff you do not control: forum posts, third-party articles. 3:19 So if the vast majority of our digital footprint is out of our hands, asking non-technical people to stress over architecture seems honestly a bit disproportionate. 3:28 I see where you're coming from, but I would argue the five-week delay was more about, you know, intention and prioritizing the project, not an insurmountable technical wall, though that does lead directly to the biggest pushback we hear. 3:41 Why build a personal website at all? Exactly, because millions of professionals already have LinkedIn profiles. It's... is the modern global Rolodex. Right. 3:50 The assumption is that having your meticulously documented work history there is enough to be found. Yes. Why reinvent the wheel and build digital plumbing from scratch? 3:58 Because the AI tools you want to be found by often cannot see that Rolodex. We have to understand how AI gathers data using things called crawlers. The automated programs. Yes. 4:09 Think of a crawler like an automated library assistant. Its only job is to run down the aisles of the internet, skim every open page, and build a catalog. But these automated assistants do not sign into accounts. Right. 4:20 They do not have user credentials. Exactly. So most of LinkedIn's valuable nuanced data only renders after a human logs in. If the crawler cannot unlock the glass case, it just walks right past it. 4:31 But wait, public profiles definitely exist on LinkedIn. A crawler should at least see a basic public page with a name and title, right? You would think so, but there is a second much harder technical barrier. 4:42 Every website on the internet has a file called robots.txt. Oh, I have heard of that file. Yeah. It acts as a set of strict traffic directives. 4:50 It, it's like a sign at a building's entrance telling automated systems which rooms they can enter. When Alex tested his own LinkedIn post URL with an AI tool, it hit a wall. 5:00 LinkedIn's sign actively refused entry to the automated reader. Because they wanna protect their data, I assume. Exactly. They are walled gardens. 5:07 Their business model relies on keeping users there, not letting third-party AI models scrape everything for free. So their very security makes you invisible to search. 5:16 So you are saying you just do not own the walls of that garden, which means you cannot set the rules. Right. Which is why you need a simple page you own, and then you have to use tools like Google Search Console. 5:27 What is that tool? It is like giving the post office your new address. If you don't submit a change of address form, the mail carrier has no idea you live there. 5:35 You have to actively push your sitemap so the library assistant knows to catalog your expertise. 5:40 Okay, I understand the mechanics of the plumbing, but let us test the assumption that pushing this data actually solves the problem. Large language models are, well, notorious for just making things up. 5:51 Hallucinations, yeah. Right. Let us look at the reliability. Even after Alex built the site and submitted his info, he asked Gemini about himself, and what happened? 6:01 Well, he got mostly accurate information about his thirty years in tech and his role at Microsoft. Mostly, yes. But it was accompanied by two highly specific errors. 6:09 First, it pulled up a forgotten volunteer stint, calling him a data science mentor on an ADP list. Which was definitely outdated, yes. And second, a complete guess with absolutely zero source material. 6:21 It guessed he was trilingual and spoke Spanish. He does not speak Spanish. Right. 6:25 So if you do all this heavy lifting and the AI still unearths a dusty profile from years ago or just hallucinates a language fluency out of thin air, are we not fighting a losing battle? 6:35 That is a great point, but we have to break down why those errors happen. They are actually symptoms of the exact disease this strategy cures. Large language models do not query absolute truths. 6:46 They are prediction engines. Right. They operate on statistical probability. Yes. So when there is a vacuum of clear structured data about a person, the AI's confidence drops. 6:56 In the absence of a strong signal, it grasps at anomalies. It found that old ADP List profile because, well, it was the only piece of data giving the model any mathematical weight. And the Spanish guess? Same flaw. 7:08 Without definitive facts, the predictive text simply guesses what statistically might fill the blank space. But how does setting up a plain text website stop a prediction engine from grasping at straws? 7:19 By flooding that vacuum with an overwhelming signal of self-correction. Look at the mentorship example from the text. Alex asked a broader question without using his name. "Who offers free SQL Server mentorship?" 7:30 Okay, a general search. Right. And Copilot's Auto mode just spat out generic community groups. 7:36 But when he used Copilot's Think Deeper mode completely signed out, his free mentorship program was the very first answer provided. Wait, really? With the correct details? Yes. It did not guess. 7:47 It pulled the exact duration, frequency, and target audience directly from his page, and this top-ranked result happened just five days after he told Bing about his site using Bing Webmaster Tools. 7:58 That is a very fast turnaround. Exactly. It proves that when you replace a vacuum with clear data on a platform you control, the AI uses it as the anchor. Okay. 8:08 That is a strong outcome for an individual side project, but let us escalate this to a corporate scale because the stakes change entirely. Sure. The corporate example in the post is actually quite concerning. 8:18 Alex tested this inside Microsoft's own ecosystem. He asked Copilot's Think Deeper mode if Azure SQL Hyperscale was a good fit for a small database. Copilot said, and I quote, "Usually not by default." 8:30 Which sounds bad on the surface, I know. It is a nightmare for a business. If an AI tells a potential client your product is usually not by default a good fit, that is a catastrophic failure of messaging. 8:41 How could a small business avoid setting a trap like that for themselves? I understand the alarm, but this brings us to the most crucial lesson in the entire post. The AI did not make a mistake. 8:51 It advised the user against using the product. How is that not a mistake? Because Copilot based its answer strictly on the text it found in the official documentation. 9:01 You see, when humans read technical manuals, we apply nuance and context. AI does not have human intuition. It reads text as flat data points. So the documentation itself had conflicting information? Yes. 9:14 The text it ingested simply did not support current positioning. And when Alex shared this with the product team, it was a totally calm discovery. 9:22 They saw the exact same answers, and the wording is currently being reviewed. There is zero blame on the people who wrote it. Right. Documentation evolves over time. Exactly. 9:31 But the universal lesson here is that AI is a mirror. It does not infer your marketing strategy. It reflects exactly what you publish. 9:38 Every business has to review their own pages to make sure the message the AI is reading is the correct one. I can concede that logic. 9:46 AI acts as an unblinking mirror, but, you know, conceding that just amplifies my original concern about the burden. How so? 9:53 Now we are telling a busy local logistics manager that they have to conduct a forensic audit of every sentence they have ever published because an AI lacks intuition. And we cannot forget Marina's statistic. 10:04 Eighty-two percent of what AI cites is earned media. The external stuff. Right. 10:08 Even if a business owner perfectly sanitizes their own website, the vast majority of the data feeding the AI comes from sources they absolutely cannot control. The burden of auditing is real. I will give you that. 10:19 It is a necessary shift in responsibility. But regarding earned media, yes, you cannot control third-party forums. 10:26 However, without your own media acting as an anchor, the AI has no baseline truth to weigh against that external chatter. So your controlled data stops it from drifting. Exactly. 10:35 It prevents the engine from running away with whatever random artifact it finds. But look, I think we can find some common ground here. Let us hope so. 10:44 Whether a listener is ready to build a website tomorrow or still feels intimidated, we can both agree that burying our heads in the sand isn't an option. Testing your current AI visibility is a mandatory first step. 10:54 I definitely agree with that. You have to know where you stand. It's a reality check. 10:58 And if there is one golden rule to conducting that test, it is this: You absolutely must use a private window or be completely signed out. The mechanics there are vital. 11:07 If you search while logged in, your browser holds cookies and your local history. The AI uses that local context to skew its answers, basically feeding you what it thinks you want to see. Precisely. 11:17 Remember that funny moment in the text, Alex tested a search on his wife's phone while she was signed into her account? Oh, right. He asked for free SQL Server mentorship? Yeah. 11:28 And Gemini Pro told her the most accessible mentorship was actually right in her own home. It recognized her husband's role because it understood her local environment. It bypassed the global internet completely. It did. 11:39 It was not objectively ranking his webpage at all. That is a brilliant illustration of why you have to emulate a stranger. You have to strip away that local context. 11:48 I still maintain, though, that for the busy non-tech professional, managing all this and auditing legacy text is a heavy lift, especially with so much earned media out there. 11:56 And I maintain that taking control of your baseline facts is just a non-negotiable professional requirement today. 12:02 If you do not define your parameters, the AI will define you using an outdated volunteer profile or just leave you out entirely. 12:08 Well, for the listeners who want to see exactly where they stand today, we can offer one immediate practical action based on the post. You do not need to architect a website this afternoon. Right. Just a quick test. 12:20 Exactly. Simply open a private incognito browser window right now. Go to your preferred AI tool and ask it a straightforward question. Who offers what you do in your city or online? Review what the AI produces. 12:32 Is it accurate? Is it relying on a decade-old profile, or is it just guessing entirely? It is a rapid diagnostic test. 12:39 And for those wondering about the actual technical steps to build visibility, the post notes that a step-by-step guide is actually coming in a Friday tip from the author. 12:48 It really operates as a fascinating thought experiment that becomes a real-world audit very quickly. It truly does. 12:54 We invite all our listeners to explore the mechanics we discussed today by reading the full post entirely for free at intecornot.com. That is spelled out as intecornot.com. 13:05 As we close today, think back to that automated library assistant walking down the aisles of the internet right now. 13:11 When it walks past your aisle, is it gonna find a clear, readable sign, or is it gonna hit a locked door and walk away? Thanks for joining us on the debate. We will see you next time.