Christian Johnson with Data Cold Storage and the Impact of the LLM in AI – HGG579

Christian Johnson is back and we spend some time catching up on Maple Grove Partners. We have a discussion on the benefits of cold storage is how to make the most of it. We then discuss the increasing accessibility and affordability of language learning models (LLMs) for both enterprises and individuals. Is the future where everyone will have their own language learning model, with some being cloud-hosted and others privately hosted? Christian predicts that the availability of LLMs will further accelerate their impact on daily life. They anticipate that LLMs will become more mainstream and affordable, fueling an already growing trend and making previously inaccessible things more attainable.
Thanks for listening! Links and show notes below.


Full show notes, transcriptions (available on request), audio and video at

Join Jim Collison / @jcollison for show #579 of Home Gadget Geeks brought to you by the Average Guy Network.


Join us for the show live each Thursday at 8pmC/9E/1UTC at

Popular Tags: Podcast, Home Gadget Geeks, Jim Collison, Christian Johnson, client-side platform, email capabilities, managed model, self-hosted instances, data centers, Google, hosting, email, small businesses, Gmail accounts, mailbox, spam, professional email address, garbage email address, chat GPT, Training Model, model, cold storage


Find Us!

Join us in the Facebook group at

On Discord at

Save as much as $110 on your first 5 boxes of HelloFresh

Use the code MQU328 and get $40 off any machine purchase and a free capsule dispenser. If you are drinking bad coffee – STOP. Great coffee, one cup at a time from people who really know coffee. Check them out at and use code MQU328 and get $40 off your first machine!

Show Segments:
[00:02:31] $10 hosting plans.

[00:04:40] West Coast server expansion.

[00:08:38] Gmail as a garbage email address.

[00:11:34] Cold storage and data security.

[00:16:06] Storage options for home users.

[00:20:49] Querying Cold Storage Data.

[00:25:41] Local transcription and search.

[00:27:22] Streaming Wars and Cable TV.

[00:32:16] Changing consumer streaming landscape.

[00:35:38] VPNs and internet privacy.

[00:39:18] VPNs and data privacy.

[00:45:11] AI and large language models.

[00:49:18] AI and deepfakes.

[00:53:30] Impact of AI on daily life.

[00:55:12] Acceleration in AI job market.

[00:59:31] AI and automation in business.

[01:05:02] Limitations of AI advancements.

[01:11:26] Homogeneous state and error.

[01:14:20] Commodity programming and chat GPT.

[01:17:35] Human-like errors in AI.

[01:23:06] Trust and accountability in AI.

[01:26:45] Changing job landscape with AI.

[01:29:00] Join our Discord community.



More from Chat GPT: Notable Quotes from the Show

[00:00:08] – “Here on Home Gadget Geeks, we cover all the favorite tech gadgets that find their way into your home and other places”
[00:04:48] – “Part of our think big strategies have been, how can we kind of do what I call micro Maple Groves in various locations to get latency East and West and get better content distribution?”
[00:07:41] – “I can’t tell you how many small businesses I see where they send you a business card and it says at and like your, your thought about the maturity of the business instantly goes down a notch because it’s like, you know, why? Like if you’re if you’re that invested and interested in running the business seriously, like, why don’t you have your own name?”
[00:09:08] – “I don’t see Gmail as a less professional email address. But I do, I do, I get a sense it’s a garbage email address.”
[00:11:24] – “but it also helps. You can think of it kind of as like a vault, right?”
[00:16:26] – “And that’s great. But it’s actually not average guy, because doing it yourself at home and meeting the same level of durability that usually these companies are promising in their SLAs for these types of services that you’re paying 15 bucks a month for, pretty hard to do by yourself, less so about the initial build and much more so about the upkeeping maintenance, right?”
[00:19:11] – “if I didn’t have high confidence in the provider and their security integrity trust model, every single bit, I would write to that cloud, I would client side encrypt myself just because I’m a tinfoil hat.”
[00:21:56] – “I would say ability to retrieve and use the data is something that I think is oftentimes an afterthought.”
[00:25:18] – “Just one of the things you mentioned that made me think was that, you know, well, some of my podcasts I transcribed and some of them I didn’t.”
[00:26:09] – “But you know, how cool intuitive, like, how hard is it that we can’t just offer out of the box, like advanced search to like I shouldn’t have to like, what podcast was that? And did I tag it right? And like, no, I should be able to search a keyword and all my multimedia should be transcribed.”
[00:29:03] – “I mean, Microsoft has pretty much openly come out and said that Windows 12 will be the AI operating system, which is code for we’re moving back to, you know, dumb, dumb thin clients for end users, and all the big compute will be in the cloud. And you’ll pay as you go for the different services that you want to make some of those things that just like, if there were an iterative product where there wasn’t this concept of digital licenses, and people paid for software out of a box like they used to, you probably would see a lot of these features in the operating system by now. And so it’s just one of the sadder realities, I think of software as a service is like industry realized like, oh, we make way more money if we’re doing SaaS than we do if like, you buy a you buy a box and you love it for 20 years, and then you call us every two weeks for support, and patches and fixes, and we don’t see a dime.”
[00:31:40] – “As we pay $73 a month, to watch the Big Bang Theory reruns, right?”
[00:35:38] – “The thing that’s interesting about VPNs is I think of them as gateways to all these different versions of the internet and how it represents itself, right?”
[00:37:58] – “I think they’re smart. I think they’re pretty cheap at this point. So if if you are trying to figure out like, gee, there’s this $5 a month that I just really want to spend, but I don’t know what to spend it on. Like, probably not a bad option.”
[00:42:31] – “And I don’t think I don’t think the level of effort is nearly as painful as maybe it was in ages past.”
[00:44:01] – “I really, there is just no matter how much a company wants to tell you there is a magic reset button on the internet, there isn’t. Don’t be fooled.”
[00:48:21] – “Ah, yes, Watson is now dead.”
[00:51:40] – “I do think we’re approaching things that will mimic general artificial intelligence much quicker than maybe people anticipated.”
[00:54:29] – “No, no, straight, it’s an excellent point that really can’t be underscored enough is that one, the guys who got involved in OpenAI and like, yeah, Musk had some involvement and Microsoft, obviously, big investor, etc, etc. But like at the end of the day, the folks that came up with, yeah, we’re just gonna make it available to the masses and see what happens, like, genius, because don’t worry, they’re not going, they’re not going home at night, poor and impoverished, doing charity work. And yet, they’ve, they’ve made a fairly obscure technology for the general population, very, very available, very, very rapidly. And that is definitely a huge part of, I think of what has caused this acceleration. So that point in particular can’t be underscored enough.”
[00:57:59] – “how can I put these three or four AI tools together and literally start printing money into bank accounts.”
[01:01:29] – “Which like, I’m not trivializing, but the fact that AI what took 80% of the legwork out of the equation, and there’s going to be side hustles already are just showing all up.”
[01:05:02] – “Yeah, it was Watson chat GPT too early.”
[01:06:34] – “I have all this content of me speaking that could easily be transcribed, even a crappy transcription, get it pretty close. Then it would run that model against me to really start, for me, I could start saying, what would I say, right?”
[01:08:30] – “wait, is that Jim actually on the other side of that meeting camera? Or did he put the stunt double up today?”
[01:13:08] – “But I find a lot of times if you put it down a path that you don’t like, you can get you can put reinforcement learning back into that model just by redirecting its line of thinking and questioning and augmenting it with your own kind of human correction.”
[01:14:20] – “And so it’s generative capability applied to those specific domains where you do have reasonable control over quality checking and independently assessing and verifying.”
[01:19:18] – “We allow humans to be wrong. Oh, that’s human, right? AI, it’s like, Oh, that machine got one thing wrong. We can’t trust it.”
[01:23:06] – “But we, we, we miss this fundamental concept of trust in it.”
[01:24:52] – “And I think we might find that even if one-to-one replication of form is a bit elusive, if the function is the same at the end of the day, eh.”
[01:27:12] – “Not yet, but ask me in 10 years.”


Find Us!

Join us in the Facebook group at

On Discord at

Get the Home Gadget Geeks subscribe links at is powered by Maplegrove Partners web hosting. Get secure, reliable, high-speed hosting from people you know and trust.  For more information visit