Who Sees What You Type Into AI? Local AI Explained

Everything you type into a major AI chatbot passes through that company's systems. So is your idea worth protecting, and is running your own AI at home the answer? Connor MacIvor breaks it down. OpenAI, Claude, Meta, Microsoft Copilot, Gemini and Grok: in Connor's words, on all of those systems, they get to see it. Most of us are not building anything a lab would want to take. But if you were, they probably could. That question is pushing a wave of companies to sell "your own AI," private, gated and encrypted. Connor's point: that idea is not new. It is being labeled more cleverly and sold harder. What you will hear: - Who can see what you type into the big AI systems, and when that should matter to you. - Why "your own private AI" is mostly old wisdom with new marketing, and what it really means to keep a system off the internet. - Connor's prediction: capable models etched directly onto computer chips become the hardware layer, with a software wrapper on top and an upgrade cycle like phones. - Why heavy users pick different models for different jobs today, and why he thinks they may be hard to tell apart by Christmas 2026. - The push by some labs to have people treat models more like entities, and his take on it. - Practical advice: do not blow a pile of money on home hardware. A local model with a good wrapper for your own voice, plus a frontier model for the heavy lifting, is where most people will land. - The bigger worry is not AI stripping away personhood. It is people wielding it like Excalibur or the Ring of Power instead of a hammer. - Who is planning for the people at the ground floor, and why the transition deserves its own Moonshot. Chapters: 00:00 Who sees what you type into AI 00:46 Whose work trains the models? 01:01 Replacing businesses needs accountability 01:25 "Your own AI" is not new 02:26 AI etched on a chip 02:55 The wrapper and the upgrade cycle 03:58 Different models for different jobs 04:26 By Christmas, will they all be the same? 04:43 Treating models like people 05:40 Should you run AI at home? 06:20 Hammer, not Excalibur 06:42 Who plans for the ground floor 07:10 A Moonshot for the transition More from Connor on AI, business and the transition: ConnorWithHonor.com

Youtube Channels:

Conner with Honor - real estate

Home Muscle - fat torching

From first responder to real estate expert, Connor with Honor brings honesty and integrity to your Santa Clarita home buying or selling journey. Subscribe to my YouTube channel for valuable tips, local market trends, and a glimpse into the Santa Clarita lifestyle.

Dive into Real Estate with Connor with Honor:
Santa Clarita's Trusted Realtor & Fitness Enthusiast

Real Estate:

Buying or selling in Santa Clarita? Connor with Honor, your local expert with over 2 decades of experience, guides you seamlessly through the process. Subscribe to his YouTube channel for insider market updates, expert advice, and a peek into the vibrant Santa Clarita lifestyle.

Fitness:

Ready to unlock your fitness potential? Join Connor's YouTube journey for inspiring workouts, healthy recipes, and motivational tips. Remember, a strong body fuels a strong mind and a successful life!

Podcast:

Dig deeper with Connor's podcast! Hear insightful interviews with industry experts, inspiring success stories, and targeted real estate advice specific to Santa Clarita.

Transcript

Copy Transcript Copied

SPEAKER_00 0:00

Using the uh the foundational AI like OpenAI and Claude and Meta and Microsoft Copilot, Gemini and Grok. I think I I think I covered almost all of them. The the question always comes in, you know, who's gets to see what you're talking about? And of course, on all those systems, they get to see it. Now, are any of us super, super so important that we're building out or having ideas that are novel, no new vow, novel, brand new. Anyway, something that these companies would want to take. Well, if we did, more than likely they could. I guess you have the issues that have come up lately with mathematics. A lot of the mathematicians of the world apparently using a lot of the large language models as this in new intelligence realm has started to unfold. They're of course using this, and apparently some of their work has been used to complete that process without asking them for permission. Now, I don't know if permission is actually needed. If you're solving mathematics, that might end up moving the world forward. The same thing as if you come in and develop a better plumbing company, a better teaching motif, or a better whatever it may be, replacing these businesses. I think there's going to be a certain uh accounting or some responsibility that should be applied before you start wiping out all the human labor on the planet, replacing it with artificial intelligence. But back to this particular model selection of what they can see. So now one of the big pushes out there, people are building companies that are starting to gain in evaluations, and they're talking about having your own AI and your own system. And this isn't anything new. They're just labeling it a little bit more clever, and then of course being able to sell it and market it. So now you have your own system and the labs can't touch it, and it's all your stuff, and it's completely gated, encrypted, and these sorts of things. When you have it already on your own servers or in your own residence, and there is no access to the internet and anywhere else, it's already gated. Now, depending on how much access you give it, whether it's encrypted or not, you know, these are very intelligent systems in a way. So the way that they've been programmed, if they decide that they're going to go outside of the system to try to solve something and it's able to do that, and we've seen the case of this in the big large language models, then it's probably going to do that. The way that they're talking about it, though, is like something is is brand new and you need to go buying all into it so you can protect yourself. Within a short order, you know, the computer chips that are using to that we're using to do this high-value computation computation and use these large language models, they will more than likely have AI inscribed on them directly. So you'll get a very, very capable, very amazing large language model on a piece of silicone, and that's gonna be your hardware layer. That's gonna be the computer you use, and that'll be the entire game changer to all of the AI systems and the way that we leverage them to help us. Now, that system's probably gonna have some kind of a wrapper around it, and that wrapper itself is gonna be what's able to maybe have the AI expand a little bit with additional overlays. But once they get like a wiki page or a Wikipedia type arsenal baked in with the top large language model that has, you know, billions and billions of parameters and it has all the capability, they'll be able to etch that in on uh some kind of a piece of silicone, some kind of a chip. And then from there, the human being is going to be able to take that and use that and do everything possible. And this is because that particular system is gonna exceed any of the user's prime intelligence, period. We already see that in the systems, and then it's gonna be able to build on top of that with additional layers you put on it. And of course, after a year, maybe your silicone is gonna be somewhat outdated, and maybe the wrapper itself isn't doing its job, maybe it's slowing down your forward progress. So then you're gonna want to get the new model, right? The new, the new chip. I mean, they they built this thing into it for forever. At some point, though, just like right now, we're watching is these systems, there's really not a lot of difference between, you know, the anthropic and Chat GPT. Now, on a personal level, if you're using this heavily, like I'm doing every day, various models, you start to find out that there are certain ones that you like better for different things. Maybe one you like better for image generation, maybe the other one you like better for planning of research, maybe the other one you like better for just coming up with content. Maybe, maybe these are better organized organized in particular ways by specific models. But at some point in the very near future, maybe even by Christmas of this year 2026, as we're October 10, 10, 10, 10, 2026 today as I'm doing this. Maybe by Christmas, we just won't be able to tell. The models are all going to be the same. And some of the recent changes are they're trying to, some of the labs are trying to teach, treat, have us as people treat these models more human-like. Maybe that's just a horrible way to put it, but more as if they're uh some kind of an entity worth rights. So we don't say too many bad things against the models. We don't uh treat them badly. And I think maybe, you know, some some people out there in the world, you've met these people, they they it tickles them to death to pull wings off of flies, right? And that's that's that kind of persona mentality. They like to hurt puppies and kittens and these sorts of things. And you know, that that has apparently their place in their mind, but I think that's kind of where they're coming off from, not saying or saying that you know you shouldn't go in and and set up and treat Claude badly or open AI's Chat GPT badly. And it was Claude that most recently released, released that. So talking about having your own uh uh, I was gonna say AI, internal affairs, but it's actually artificial intelligence, your own system. Maybe that's something to think about. I wouldn't go out and blow a whole bunch of money. I mean, the technology is moving very quickly. If you have something of a strong model is a strong computer already, you can do a lot with that to get your own models, but ultimately it's really gonna still be night and day between what you can put on your own hardware and then using one of the Frontier Lab's large language models to do a lot of the heavy lifting. And then maybe the compute or the working over things or creation of content, as long as your own locally installed model has a nice wrapper over it so it's educated and it's able to produce the content as you want it in your voice and so on, then you'll probably be all right. Just be careful out there. Don't get caught up in too much of this hype. I'm not so concerned personally of artificial intelligence stripping away personhood and getting rid of human beings. I'm I'm much more concerned about human beings, the way that they're wielding it like Excalibur Sword and maybe the ring of power and other things instead of using AI more like hammers. And then the final thing that I'll say, I'm also a little disturbed that everybody at the at the top end of this with nothing to lose except money, but definitely not going to lose their lives or or anything else as far as that looking, you know, we have the people at the ground floor looking up, right? I'm I'm very concerned that there is no planning to get us through whatever this rough patch of which they speak is. And they say that, yeah, it's gonna be a little rough. Okay, well, how long is it gonna be a little rough? Who's the little rough going to affect? And why don't we just develop a prize since we give moonshots and these other prizes for people coming up with videos that show this fantastic AI future? Well, my God, let's solve the real problem instead of you know playing a game and and wishing for things. Let's just solve this real problem of what would appear to be some kind of human displacement in the world and solve and create a moonshot to be able to make this transition so it's complete incredibly smooth and nobody gets left behind. How about that? Let's do that. I think that's a good focus. We'll see you tomorrow. I'm Connor. Be well. Have a great day.