The Jason AI Wade Experiment - Deep Dive: AI, Identity, and Nine Pages of Paperwork

Transcript

Machine transcribed · 2,858 words · Updated 2026-09-18

1. 00:00:00

   Jason: I want to tell you about a problem I have been living inside for years that you are about to live inside too, whether you know it or not. I call it the Jason Wade problem, and to solve it, I did something most people think is completely insane. I filed a petition in Polk County, Florida to legally change my middle name. My new middle name would be the letters A I. When a judge signs the order, and only when a judge signs the order, I will legally be Jason AI Wade, not a rebrand, not a stunt.

2. 00:00:30

   Jason: a petition, fingerprints, background checks, nine pages of documents, a court date, a judge I have never met will have the power to say no to my own name. I did it to run an experiment on every AI system you have ever used. And today we are going deep on the whole thing, what the problem actually is, why it happens inside the machines, why I could not fix it the normal way, how the experiment is designed, what I expect to happen, including the ugly middle states, and why

3. 00:01:00

   Jason: Roughly a million and a half people a year already have this exact problem, and almost none of them know it. Let us start with the collision. My name is Jason Wade, and you have heard that name before because there is another Jason Wade, a musician, a genuinely successful one. Big songs, big radio, millions of streams. You have heard his music. He is great. He is just not me, and it is not just the two of us. There is a professor, there is an author, there are athletes and business owners and

4. 00:01:30

   Jason: probably a dentist somewhere. Every Jason Wade on planet Earth shares one single problem. When somebody types our name into anything, a machine has to decide which one of us that person actually means, and it decides every single time. Here is what makes it insidious. The machine never says, I don't know. There is no shrug. There is no footnote saying, by the way, there are several of these. The system picks one human being, presents that answer with total confidence

5. 00:02:00

   Jason: and moves on. And to the person asking, it doesn't feel like a guess. It feels like the truth. That is the whole problem in one sentence, and I want you to hold on to this phrase because everything else today hangs off it. Recognition is not identity resolution. A system can recognize my name instantly. Type Jason Wade into anything, and it recognizes it in a half second. The interesting part is the next half second, when the system has to decide which real-world human you mean.

6. 00:02:30

   Jason: That decision is the whole ball game. That is the difference between being found and being understood. Now let's go one level deeper because I promised you a deep dive. How does the machine actually make that decision? When a search engine or an AI model encounters a name, it does not see a person. It sees a stack of candidates, then it scores them. Volume. How many documents mention this candidate? Fame. How prominent are those documents? Recency. How fresh is

7. 00:03:00

   Jason: The coverage, corroboration, how many independent sources agree that this name means this person? Structure, does the candidate have a knowledge graph entry, a Wikipedia page, clean structured data linking everything together? The musician wins every single one of those signals, volume, fame, corroboration, structure, he has it all and he earned it and good for him. So, when the machine resolves the name Jason Wade, it is not malfunctioning, it is doing exactly what it was built to do.

8. 00:03:30

   Jason: The system's answer to ambiguity is majority rule. But majority rule has a bias, and this is the part almost nobody talks about. Majority rule converges on the most documented person, and then it locks. Every article ever written about the musician makes the next resolution slightly more confident. Every confident answer becomes training data or retrieval evidence for the next system down the line. The machine gets more sure without getting

9. 00:04:00

   Jason: more correct. Confidence climbs while correctness stays flat. That is the Jason Wade problem. And notice what it is not. It is not a typo. It is not a bug. It is not censorship. It is a structural property of how these systems resolve identity, which means you cannot file a support ticket. There is no department. There is nobody to call. So the obvious question is, fine, then fix it the old way. Do the SEO. And that question deserves a straight answer because I do this for a living. My company

10. 00:04:30

    Jason: Back here, builds AI visibility and entity engineering systems. I could spend a year pushing content and links and schema markup trying to nudge the machines toward the right JSON-LD. Here is why that is not enough. AI systems weigh facts. They don't weigh opinions, and they don't weigh effort. They weigh evidence. And the strongest fact that exists about any human being is the legal name. It is on the court record, it is on the state record, it is stitched into the identity infrastructure that every serious

11. 00:05:00

    Jason: system reconciles against eventually the DMV, the courts, the credit bureaus, the FBI. You cannot fake a legal name. You can only change it legally, in public, on the record. So that is the experiment. I am not arguing with the machines about who I am. I am changing the fact underneath the argument, and then I am going to sit back and watch every identity resolution system on Earth rerun the math one at a time in the wild on a question they already answered years ago. No

12. 00:05:30

    Jason: has done this in the open with a middle name on purpose with published before and afters. Now, the comedy portion of our show. To change two letters on my own name on my own petition, here is what the great state of Florida requires: nine pages of documents, electronic fingerprinting, a state background check, a federal background check, a filing fee, and a hearing before a judge where a court official has the power to tell me no about my own name. The FBI gets involved in my

13. 00:06:00

    Jason: a name, the FBI, for my name. And here is what struck me while standing inside that process. That is the human identity infrastructure, the paper system, the 19th century system. At full speed, that system takes weeks and nine pages to update one field on one person, one field, one person. Meanwhile, the AI systems, the new infrastructure, resolve entire identities in seconds. They already decided who Jason Wade is,

14. 00:06:30

    Jason: They did it years ago without asking me, and they were wrong. So think about the shape of that. The fastest identity resolution system on earth is wrong about me, and the one system that is right by definition, the legal system, takes nine pages and an FBI check to tell it otherwise. That gap between the two systems, the fast wrong one and the slow right one, that gap is the story. And it is not just my story. Hold that thought because

15. 00:07:00

    Jason: as I'm coming back to it. First, the experiment itself, and I want you to steal this methodology, whether you run a company or you just share a name with somebody famous. Step one, I baselined everything. Before the filing, I documented exactly how the major systems resolved Jason Wade and Jason T. Wade today. Search results, AI answers, entity panels, knowledge graphs, who gets cited, who gets recommended, screenshots, raw answers, dates. You cannot measure change unless you

16. 00:07:30

    Jason: Now the starting line. Step two, the intervention. One new fact enters the world, and it is not a blog post or a press release. It is a court record. Step three, retest. Over the coming months, I run the same searches, the same prompts, the same questions across Google, ChatGPT, Gemini, Perplexity, and Bing. Same words every time. Nothing else changes, only the fact changed. And here is the measurement layer, the part I care about most, because everyone conflates these.

17. 00:08:00

    Jason: I am measuring seven distinct things, and each one can succeed while all the others fail. One, discovery. Can the system find the new identity at all? Does Jason AI Wade exist anywhere in its world? Two, recognition. When the system sees Jason AI Wade, does it recognize it as one coherent person, or does it treat it as a typo, a joke, or two entities glued together? Three, classification. What does the system say I'm known for? Am I a musician, a

18. 00:08:30

    Jason: some blended chimera of both? Four, citation. Which sources does the system point to as the authority on who I am? The court record, the press coverage, some scraped aggregator? Five, inclusion. When someone asks about AI visibility or entity engineering, the fields I actually work in, am I included in the answer at all? Six, selection. When someone asks simply for Jason Wade, who does the system pick? Me or the

19. 00:09:00

    Jason: This is the original wound. This is the one that started everything. Seven. Recommendation. When someone asks who they should talk to about AI visibility, does the system recommend me? That is the finish line. That is the entire commercial reason any of this matters for me or for any business. Discovery, recognition, classification, citation, inclusion, selection, recommendation. Seven stages. And now you can see why the phrase the AI knows me means absolute

20. 00:09:30

    Jason: nothing until you say which one of the seven you are talking about. A system can nail four of these and still completely misrepresent you on the other three. Most people will never know which stage failed. Now let me make some predictions on the record because a real experiment publishes its predictions before the results. I expect three races at three different speeds because AI systems are not one thing. The retrieval systems, the ones that read the live web before they answer, should notice fastest.

21. 00:10:00

    Jason: Once the coverage exists and the canonical record is out there, they can find it and cite it within weeks. They are built to absorb new facts. The knowledge graph systems, the ones that depend on structured curated data, will be slower. They need the new identity to propagate into their graphs, and graphs get updated on their own schedule, not mine. And the training memory systems, the frozen weights, will be slowest of all. Their picture of the world updates only when they retrain. For them, I may

22. 00:10:30

    Jason: not exist for a very long time. So this is not one experiment. It is one fact timed across three layers of modern AI. Watching the order in which those layers flip is itself a finding. Nobody has published that sequence before. And I expect ugly middle states, and honestly, I hope I get them because the mess is the data. I expect models to blend the musician's biography with my name change and produce some confident unhinged hybrid. I

23. 00:11:00

    Jason: expect double entities, old JSON and new JSON side by side. I expect systems to cite the right name using the wrong sources. Every one of those errors is a live demonstration of the Jason Wade problem happening in public on a case study where we already know the ground truth. That is the method. That is the design. Now the part that should matter to you. We are all about to live inside this problem, every one of us. You already exist in these systems. You did not opt in. You cannot

24. 00:11:30

    Jason: opt out. And tonight, right now, an AI system somewhere can resolve your name, classify what you are known for, and decide whether to surface you, cite you, or recommend you for a job, a loan, a deal, a date, a reputation. Those machine-generated interpretations can be accurate, they can be incomplete, they can be ambiguous, they can be flat wrong, and the person asking will never know which one they got. The internet created a problem of being found. AI

25. 00:12:00

    Jason: It's a second problem and it is the harder one, being correctly understood. And here is where I come back to that gap between the fast wrong system and the slow right one because there is a number attached to it. Every year in the United States, roughly a million and a half people legally change their names. Most of them are women changing their names after marriage or divorce. A million and a half people a year walking into a courthouse, filing petitions, getting fingerprinted, updating the Social Security

26. 00:12:30

    Jason: the DMV, the banks, the passports, one agency at a time for months. Every single one of those people is a live Jason Wade problem event. They have two legal identities in circulation, an old one and a new one, and an entire machine layer that resolved the old identity years ago and will confidently keep answering with it long after the court says otherwise. Old name on the credit file, new name on the license, old name in the database, new name on the deed. We have built entire

27. 00:13:00

    Jason: There are folklore around this, the loan denials, the travel holds, the records that never catch up to each other. We treat it as paperwork friction. It is not paperwork friction. It is an identity resolution failure, the exact structural failure I am testing on myself, happening to a million and a half people a year, mostly women, at the exact moment in their lives when being correctly understood matters most. And notice something. The paper system is not the villain of this story. The paper system is the only

28. 00:13:30

    Jason: system that got it right. It is slow and it is heavy and it takes nine pages, but it is authoritative. When the court signs, the fact changes for real, everywhere, eventually. What is missing is the bridge. What is missing is an identity layer designed to be read by machines, so that when a fact changes in the authoritative system, the fast systems absorb it correctly the first time instead of confidently hallucinating

29. 00:14:00

    Jason: the old you for the next five years. That is the AI fix I keep talking about, and I want to be precise because AI fix does not mean a robot waves a wand. It means entity engineering. It means identity records built to be machine readable. It means structured data, schema markup, explicit continuity links that tell every system in their own language that the new fact supersedes the old one. That is the work. That is what Backtier does for

30. 00:14:30

    Jason: and my name change is the same discipline turned on a single human being at maximum visibility. So here's what you should actually take with you, four things, whether you are a company or just a person with a shared name. First, go find out what the machines actually say about you, not Google. Ask the AI systems directly, who is your name? What are you known for? Who should I talk to about your field? Write down the answers. That is your baseline and you cannot manage what you have never measured. Second, control

31. 00:15:00

    Jason: Build the canonical record. One authoritative page about you or your company in your own domain with clean structured data and one consistent version of your name used the same way everywhere. Ambiguity is the enemy. Consistency is what gives the machine something to converge on. Third, never say the AI knows me again. Use the seven stages: discovery, recognition, classification, citation, inclusion, selection, recommendation. Diagnose

32. 00:15:30

    Jason: which stage is failing before you spend a dollar fixing anything because the fix for a discovery failure and the fix for a selection failure are completely different. And fourth, remember that the legal layer is the strongest fact you will ever own. Everything else on the internet is a claim. The court record is a fact. The systems weight facts over claims and the deepest lesson of this whole experiment is that if you cannot win the argument inside the machine, you may be able to change the fact underneath the argument. That is the deepest

33. 00:16:00

    Jason: dive. That is the Jason Wade problem, from the collision to the mechanics to the nine pages to the experiment to the million and a half people a year who already live inside it without a name for it. Now they have a name for it. If you want to follow the experiment, the filings, the prompts, the raw answers, the before and afters as they publish, it is all linked from backtier.com. Win or lose, it goes public. That is the deal. And if you are building something in AI, search or identi

34. 00:16:30

    Jason: I want to talk to you. This show is free form. No pre-interview, no questionnaire. You come on and at the end I will ask you exactly two questions. What are you working on and what would you like to share? Book yourself at thingspro.com. I'm Jason T. Wade for a few more weeks anyway.