"I wonder if we are at the point where you can load an AI client in an old phone?

That would be pretty powerful coupled with 11gbs of source prepper data to search and use within seconds.

That's always been the one limiting factor about all those prepper files. Finding what you need and using it. AI could turbo charge that into a preppers dream."

I write that in a response to a topic about e readers. And I could go ask AI right now. But for the sake of conversation. Thought I would post this.

If that's possible. Even in a tablet or laptop. That would make all those gb's of prepper data we've been hoarding really usable. 

The thing with the phone is you would probably have plenty of room to load the gbs of source data. But if not, the older phones with external SD cards would be a valuable choice. You could even load YouTube videos.

I see AI more like Cliff Claven then something useful.

Quote from: Uomo Senza Nome on November 02, 2025, 08:26:48 PMI see AI more like Cliff Claven then something useful.

Have you spent any time with it?

Every day. I ask it questions. It is invariably wrong every time in some regard. Some of the stories are told well, explanations seem legitimate. But just like Cliff Claven, very often wrong.  Usually it does the best when keeping an answer short. But it seems built to want to impress us with its detailed knowledge. That is where it stumbles.

In criminal statements it is often the details that trip people up. People often try to make something appear more credible by putting extra details into their statements when they are lying. All this does is raise red flags and make it more suspect. 

I'm not saying it is never right, just wrong way more often, 70-80% of the time, about something.

Quote from: Uomo Senza Nome on November 04, 2025, 06:46:38 AMEvery day. I ask it questions. It is invariably wrong every time in some regard. Some of the stories are told well, explanations seem legitimate. But just like Cliff Claven, very often wrong.  Usually it does the best when keeping an answer short. But it seems built to want to impress us with its detailed knowledge. That is where it stumbles.

In criminal statements it is often the details that trip people up. People often try to make something appear more credible by putting extra details into their statements when they are lying. All this does is raise red flags and make it more suspect.

I'm not saying it is never right, just wrong way more often, 70-80% of the time, about something.

I would look into your queries.

I used it recently to negotiate a new vehicle purchase in a foreign country - Phillipines. It read, analyzed and understood Phillipine consumer law, common dealer practices here, and can read an analyze any contract or agreement you screenshot and upload. It was like having 5 experienced Phillipine attorneys and a secretarial pool of 29 in my living room. Except faster. I could have not done this without it. And after using it will never sign any document again without screenshotting it and having AI analyse it. I wish I had used it when I signed my lease.

Does it make mistakes? Yes absolutely. You still have to double check it's works. But if I added the time it would take me to research and analyze a given subject. Versus having AI do it. It's easily hundreds of hours to one.

Grok for instance is excellent at locating manufacturers in Alibaba. Selling specific products. It's a terrible, giant database. Being able to identify 3 manufacturers of a gas can I was looking for today, in about 39 seconds, cut out hours of research on my own.

Limiting it's access to data like within NotebookLM or running an AI client on a set of data you control. Like prepping data would refine it even further. This takes much of the "slop" out. Random runs in information maybe not related to your focus. If it set it up on the 11gb if data I have collected in the prepper realm your own ability to do anything with that vast amount of info has been turbo charged. Can you imagine how long it would take just to catalogue all that information? A ridiculous amount of time. AI can do that in a matter of seconds. That might take you and I years. At least many many months.

The other key attribute I have found extremely useful is its ability to write. Each step thru my vehicle purchase usually included an incorrect agreement. Chatgpt could quickly point out the key inconsistencies. Which I could verify without having to over analyse the third version(?) of a document. Chatgpt would then ask if I wanted it to write a follow up email and what tone I wanted to take. It would right a page and a half email within seconds, I could adjust anything including tone or specific words or phrasing. The level of writing was far above my own. And with tweaking it could come to the axact tone I wanted to use. This is within seconds. It would have taken hours to write something this polished.

I think you might want to reaxamine your use cases and queries. To say it's wrong 70% of the time. Seems your missing something.

AI is not perfect. But perfect enough that anyone can tell most jobs will be gone in 5 years. Probably far less. Hundreds of corporations worldwide are having massive layoffs already. Once this is coupled with robotics which is already rolling out at Amazon and other large corps. The question is going to become how do we feed and house the millions of people that had previously been involved in the work economy?

According to the International Labour Organization (ILO):

Roughly 70% of jobs are mental and 30% are physical labor.

AI already reads medical scans as well as a doctor. How long before you think it takes over half of the worlds jobs?

I think we're going to see a future full of dormatory type apartment buildings with cafeterias and medical centers. Funded by large corporations or governments taxing those corporations. Just to take care of a populace that is otherwise starving and rioting.

Pretty bleak. But I don't see anyone stopping corporations from using AI. When it never gets sick, never has to go the bathroom, never sexually harrasses anyone, never needs to sleep or take a day off, and costs little to nothing compared to human time.

I think you should go back and look at AI again. I don't think your seeing it clearly.

Fair enough . This morning I asked it:

"Tell me about the safety of the Sig P320 Pistol and the US Airman who was killed?"

This is an easy slow ball since there has been a lot of current media coverage about it.


It correctly identified the airman killed, calls for it's recall, the US Air Force suspense of its use and the allegations of the pistol being defective as well as Sig denials that there was anything wrong with the pistols.

If I wanted that answer I could have gone to a gun shop and listened to people hanging around talking.

Missing from the AI story was the suicide of the Airman who killed the other Airman and the court martial of two Airmen who had written false statements and then recanted.

The real story being that the suicide airman had pulled his pistol out and pointed it at the soon to be dead Airman as a joke and the pistol "went off".  Also that the Air Force had returned the pistol to service was missing. That there was shown to be nothing wrong with the P320 after all. The Airman had fabricated the story to cover it up what was basically a negligent homicide.

Lots of articles out there mention these things but AI doesn't do the work and gives and very incomplete answer. So yep, Cliff Claven or even "gun shop guy". Take your pick.


https://taskandpurpose.com/news/airmen-guilty-statements-fe-warren-m18/

Really this was a lost opportunity to yet again go over the rules of gun safety and why it is the worst idea ever to break them. Here we have the death of two Airmen and the career destruction of another two.

Quote from: Uomo Senza Nome on November 05, 2025, 06:27:47 AMFair enough . This morning I asked it:

"Tell me about the safety of the Sig P320 Pistol and the US Airman who was killed?"

This is an easy slow ball since there has been a lot of current media coverage about it.


It correctly identified the airman killed, calls for it's recall, the US Air Force suspense of its use and the allegations of the pistol being defective as well as Sig denials that there was anything wrong with the pistols.

If I wanted that answer I could have gone to a gun shop and listened to people hanging around talking.

Missing from the AI story was the suicide of the Airman who killed the other Airman and the court martial of two Airmen who had written false statements and then recanted.

The real story being that the suicide airman had pulled his pistol out and pointed it at the soon to be dead Airman as a joke and the pistol "went off".  Also that the Air Force had returned the pistol to service was missing. That there was shown to be nothing wrong with the P320 after all. The Airman had fabricated the story to cover it up what was basically a negligent homicide.

Lots of articles out there mention these things but AI doesn't do the work and gives and very incomplete answer. So yep, Cliff Claven or even "gun shop guy". Take your pick.


https://taskandpurpose.com/news/airmen-guilty-statements-fe-warren-m18/

Really this was a lost opportunity to yet again go over the rules of gun safety and why it is the worst idea ever to break them. Here we have the death of two Airmen and the career destruction of another two.

I can tell I'm not going to convince you. Lol. And I can't say I think AI won't have a devastating affect on our society worldwide. I am honestly concerned about its missuse. And the sheer lack of forethought about safety and political indoctrination. Chatgot has already had various liberal politics programmed into it. Grok is perhaps more balanced. 

But I use neither of these tools for political research. Or anything beyond practical tasks. Like deciphering contracts. Figuring out mechical problems. Sourcing product solutions. Things of a more logicical, scientific or practical activity that would take me much more time to do manually. If I could do them at all. 

I also use different AI for different things. I'm a pragmatist. This is a tool. And if you look at it just as a tool. It's uses are powerful enough to drastically change industry and business if nothing else. 

I would not turn to AI for a nuanced opinion about gun control. It's not a person. But if you wanted more nuanced information. Say the additional facts you mention about the Sig case. That could have easily been handled with follow up questions. Follow up or refining questions are essential to it's use. To say "it missed this important thing" is sort of like admitting you haven't fully understood how to use it. Because it's only as good as the questions you ask. Like a professor sitting in your living room. If you want to know what he knows - it's all about the questions. You can't fault a professor for his answers if you are the uneducated one asking the questions. 

Not meant as a slam at all. Don't get me wrong. Just trying to point out your basing an opinion on AI not being any good. Based in an answer to a limited question you asked. Without - at least it sounds like - many follow up questions. 

But again. I understand your anymosity. AI is better than a certain lower percentage of the human populace at most of the things they do for a living. That sucks. Because industry now has a replacement for the human. The perfect slave. They don't have house, feed or pay it. And it never stops working. That is going to rearrange the world. 

I'm not saying AI is a good thing. In alot if ways it's not. But you have to be able to see it's very powerful uses. Otherwise it's kind of like saying "The telephone is bullshit.  You should get off your ass and write a letter!". Or "this internet thing is a kids play toy! Get back to work!".  Lol! 

I'm an AI neophyte too. Keep that in mind. My uses and arguments for it are lame. I'm sure. I know jack shit about its full spectrum of uses. I'm using it to buy a car, find good deals, do research. Real users are using it to write software, read brain scans, I'm an idiot by comparison. Lol. So take my words with a huge grain of salt. For and against. 

But back to the subject at hand. If I had more time right now. I'd download an AI agent. Plug it into my very loose, very unorganized, probably full of duplicates prepper data dump. And ask it some questions. 

Because that pile of data is just that - a pile. Ya. Maybe if you had a first aid problem is be able to find a few first aid manuals in that 11gbs. And find an index and a page that explains what I should do in whatever situation we are in. 

But AI could catalogue that entire collection, remove dupes, and start analysing it for any number of complex issues you might come across in a paw. How should I build a large water treatment system given the supplies I have on hand? Based in whatever documents you can find in that 11gbs? 

Can you imagine how long that would take you and I to do? 

And it's "you and I"! Lmao! We're a couple of slow, stupid humans.

Even if you had a professor of biology next to you. How long would it take him to sift thru 11gb of data? Analyze it, read it, develop ideas and come up with a solid plan that is better than most? Days, weeks, months? 

AI could do that in a matter of minutes at most. Further refining it with additional queries and tweaks. You might have a very doable system within a few hours.  

Imagine all the other tasks just like that? Imagine greater databases of scientific information that you might be able to give it access too?

The difference between my old cellphone holding an SD card of prepper data. Versus a laptop running an AI agent and just that same 11gb of data is monumentally more effective as a tool after a major disaster. 

Forget about how good it is or how bad it is for society. A prepper AI seems easily within our grasp. And light-years ahead of saving pdf's. 

Don't you agree?

I wasn't asking a gun control question. I was asking whether a certain gun was safe and specifically in regards to a particular case. It provided a wrong answer. There isn't much else to talk about. Except that AI always gives wrong answers. 

If I didn't already know the answer How would I know I needed to ask it more questions? If I were dumb and lazy I would run with whatever it told me. If I needed to ask it a bunch of questions I may as well do my own research. The truth is I know how to be accurate and concise and AI doesn't and isn't. Therefore it shouldn't be relied on. 

Just because there is large volume of room for improvement doesn't mean it will always suck.  But right now? I deem it very unreliable. I'm not the only one either. I work in education environment and the teacher always knows when AI was used for a product. They usually get very low grades. Maybe at some future date things will improve. Till then it is just entertaining and that is as far as it goes.

I recently had Haven Survival reach out about their offline AI for iOS or Android which I'm currently trying out. It's an offline AI companion (with Plant and Environment scanning in the pro version) which also allows you to download offline maps, use Bluetooth to communicate via text with devices nearby, has an offline library with a handful of illustrated guides on different topics as well as a few different religious texts (Bible, Quran, Torah, etc.) and supply tracking. It's all offline after the initial download. 

Many of the features are free but you can also upgrade to the pro version for a one time payment of $24.99 for lifetime access (I have a 15% coupon but not sure if I am allowed to share as I get a kick back if it is used). I tend to avoid AI but thought it was interesting enough to test out. Really enjoying the books in the survival library so far, haven't asked the AI a ton of questions but the responses seem to be pretty reasonable and practical. 

We have approved AI tools at work that I use and they're extraordinarily helpful.  And I use ChatGPT, CoPilot and Claude at home all the time.  Last year, I asked ChatGPT to develop a curated itinerary for a 2+ week vacation to the Balkan countries with an emphais on physical activity and cultural interaction, with detailed specifics on points of interest, activities, dining options, local transporation and accommodations.  I ultimately went with an established, highly-regarded small group tour operator that specializes in exactly that kind vacation.  But the itinerary that ChatGPT developed for me in less than 2 seconds was scarily close to what the tour company offered.  It was incredibly on point and accurate.

We had a live, unscripted demonstration from an AI vendor at work and one of my colleagues asked it to develop a comprehensive privacy policy for a hosptial - but in the voice of a pirate.  We had our privacy attorney review the 20+ page document it spit out in a few seconds and he said it was dead-on accurate, aligned with applicable privacy law and extremely well-crafted.  Except it was difficult to read because he was laughing so hard. 

If you're a working professional and still reluctant to use AI: I've heard this advice from a number of professional sources: you're probably not in danger of losing your job to AI anytime soon.  But you're in imminent danger of losing your job to someone who knows how to use AI. 

Before I ever

Quote from: Uncommon EDC on April 15, 2026, 06:07:32 PMI recently had Haven Survival reach out about their offline AI for iOS or Android which I'm currently trying out. It's an offline AI companion (with Plant and Environment scanning in the pro version) which also allows you to download offline maps, use Bluetooth to communicate via text with devices nearby, has an offline library with a handful of illustrated guides on different topics as well as a few different religious texts (Bible, Quran, Torah, etc.) and supply tracking. It's all offline after the initial download.

Many of the features are free but you can also upgrade to the pro version for a one time payment of $24.99 for lifetime access (I have a 15% coupon but not sure if I am allowed to share as I get a kick back if it is used). I tend to avoid AI but thought it was interesting enough to test out. Really enjoying the books in the survival library so far, haven't asked the AI a ton of questions but the responses seem to be pretty reasonable and practical.

Before I ever paid anyone for anything like this. I would find out which stand alone AI software they were using to provide this?

Especially if they were trying to charge me. 

Gemma 4 is Google's newest family of open-source AI models, released on March 31, 2026. It uses the same technology as the Gemini 3 models but is designed to run locally on your own hardware. 

I don't see any service Haven Survival is offering except coupling their own data with a stand alone AI model. Which you can do yourself for free. 

*The ultimate use of AI in a SHTF situation is having it and your survival/post apocalyptic files downloaded to a phone, laptop or other stand alone device. 

Where you're only limited by a power source. 

Most of us have GB's of SHTF documents, manuals, books, government studies etc. I have 11gb of said data on my phone's, laptops and hard drives. Meaning one 11gb folder loaded into each.

The first thing you realize once you start digging thru that data. Is the sheer volume. And the time it would take in a SHTF scenario to develop any number of plans or strategies based on that data. Say you have a small group. And you need to build a long term water gathering and purification system. It might take a group of you several days if not weeks to digest that data, review and consider materials and access on hand, and develop the way to proceed. With AI it could develop several plans for you to consider in moments. Couple that one scenario with the many other aspects of a SHTF scenario. And you can easily see it's utility. 

AI with whatever knowledge base is already applied, coupled with the massive volume of SHTF data available. Simply loaded on an old cheap Android phone. Takes you from a group of people (or an individual) with what amounts to a massive pile of paper. To an infinitely more effective force. 

Gemma 4 is just the latest version of a stand alone AI. There are dozens. Haven Survival isn't doing anything you can't do yourself much more effectively by choosing any number of constantly evolving and improving AI. Coupled with the same SHTF data set available for free everywhere. 

I'm very excited to see what preppers do with this. Not only as a stand alone survival system. For developing the very best field options for security, food preservation, water, heating, shelter, - virtually anything you could face with whatever materials and abilities you have on hand. But to age old questions we've tried to optimize an answer to.

Just take the analysis of gear selection for instance. 75 items in your BOB, a budget, a specific area of operation, a strategic plan. Applying AIs capabilities to point out various gear choices within those 75 items. Based those key factors. Further refined by your own research into each piece of key gear. And your own knowledge what you might face. Just its help in this one situation alone is massive. Then apply it's current power to research online - find products, find information, scan every other opinion about that choice that has ever been posted online, it is light years ahead of what any one person can do. In an extreme fraction of the time. 

I'm excited to see how this effects prepping. 

@majorhavoc - I've played around with AI but nothing professionally. It's definitely a blind spot for me but am self employed so hasn't felt like a pressing area to focus. Interesting example though as I work in the healthcare space.

@Moab - Thanks! not knowing a ton about AI, that was very helpful. It let's you choose which AI model to load from a long list of compatible models. Gemma 4 is one of the options. There is also Llama 2.1 & 3.2, Hermes 3m, Dolphin 2.9.4, TinyLlama, Qwen 2 & 2.5 , Phi-3 Mini 4K, Ministral 3 3B and Dolphin 3.0. Don't really know the difference outside of their storage and memory needs so will probably switch to Gemma 4 (I loaded Dolphin3.0/Llama 3.2)

Quote from: Uncommon EDC on April 16, 2026, 12:25:05 AM@majorhavoc - I've played around with AI but nothing professionally. It's definitely a blind spot for me but am self employed so hasn't felt like a pressing area to focus. Interesting example though as I work in the healthcare space.

@Moab - Thanks! not knowing a ton about AI, that was very helpful. It let's you choose which AI model to load from a long list of compatible models. Gemma 4 is one of the options. There is also Llama 2.1 & 3.2, Hermes 3m, Dolphin 2.9.4, TinyLlama, Qwen 2 & 2.5 , Phi-3 Mini 4K, Ministral 3 3B and Dolphin 3.0. Don't really know the difference outside of their storage and memory needs so will probably switch to Gemma 4 (I loaded Dolphin3.0/Llama 3.2)

If your using Google. The first set of results at the top is Googles AI - Gemini. Chatgpt and Grok are good to. Not sure how much you've played around. But they're basically simple apps you asks questions of. Just likes google search. It's very simple. But the breadth of what it can do is huge. 

You don't need any code or fancy language. It responds just as if you were doing a normal internet search. Except it's programmed with a massive amount of knowledge and intelligence. But beyond that you can ask it to use the internet to research anything. Except not with typical page results. It will answer you like a person. With simple answers. Or you can ask for a full on research report. And continue to ask follow up questions and refine it.

At the moment I prefer Gemini. As within the three dot menu at the bottom of your answer is an option "export to doc". This exports your answer to a Google Drive Doc. One limitation of AI is it's memory is difficult to deal with. You can save projects. But it problematic. I suggest saving any kind of final product you create - a report or list of angle grinders you had it find - whatever. Save it. Getting it to give you clean data you can copy and paste is also problematic. But that feature with Gemini makes it alit easier to save your work.

Start off as if you had a professor in that area of knowledge your interested in sitting next to you. Ask it questions just like you would a normal person. How do I choose a battery station? Which ones will run the items I want to power? Which ones are the best for my application? How much do each cost? Please compare the top three and include individual specs? 

That search alone would take you the better part of an afternoon. It can answer each of those questions in seconds. It's also best to direct it where you want it to pull it's research from. I routinely ask "do current deep research into forum, group, reddit and online comments". If I want less of a technical answer and more boots on the ground answers from real people. This can helpful with everything from products to how to do something. 

But this only scratches the service. AI is an intelligent system that has absorbed a ridiculous amount of knowledge from just about every area of expertise. And can also do deep current research, make conclusions based in that research, and a write an in-depth report, letter or note pertaining to it. 

I screenshot every contract or agreement I sign overseas. Upload it. And AI gives me an analysis based in local law that is often times in local language. This provides me 95% of the capability of an attorney. For low level agreements. I could not normally afford an attorney for. Sometimes it's as simple as "is this agreement I'm about to sign the same one that was emailed to me and I reviewed yesterday". That alone might take you an hour. To compare to multipage documents. 

The list is endless.