This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
Check out internet in a box. The Wikipedia foundation supports it. IMO that is also over engineered for my needs, so built a lighter weight variant.
That's how they get wikipedia on this btw.
In my variant I have the complete English Wikipedia, a ton of official documentation on programming languages I use or might need to use, as well as calibre for eBook and pdf housing.
All running in docker containers, behind an nginx reverse proxy landing page.
Sounds like a very cool idea, the implementation.... not so much.
Also who would you use a chatbot in a survival situation when it halucinates and can be replaced with a simple search in this case?
Might be kids (under-40s) these days dont remember functional search. Or theres no easy packages available.
As long as you understand the limitations, AI is just a vastly more efficient way to find information in large knowledge bases. For a topic that you know nothing about a chatbot gives you the ability to work through the information in a conversational manner, honing in on the specific bits you're trying to understand and filling in the "things that I don't know I don't know" holes, and then you can go to the source material and verify the details.
I don't see any problem with using a chatbot to find information in a giant wiki myself.
Worrisome
Oh, do elaborate. This ought to be good.
Ask your chatbot if you can't figure it out yourself
It's as if you have a Pavlovian response to anything AI related, and then just try to deflect when pressed on your bullshit. Seems like some people on here have less mental faculties than a chatbot smdh.
Seems like some people on here have less mental faculties than a chatbot
Yeah right
I love how you just can't help yourself reply compulsively. We've already established that you can't actually make a coherent point, and are just lashing out. Take the L and go home.
I don't make arguments because you can't argue with indoctrinated people.
It's true, being indoctrinated, you're not actually capable of rational thought, and therefore unable to make a coherent point. Glad you're self aware enough to admit it.
They lost me at AI...
But yea, having the Wikipedia and offline maps stored away (e.g. for offline use on an android phone) is a great thing. I use a thumb drive instead of a interconnected thingy with "command centre". Feeling old now.
Actually, having a local agent on the system with a bunch of data on it that helps you find things would be great for most people. People really need to get over the whole knee jerk reaction to all things AI related. It's getting really tiresome.
People putting me in drawers is also getting tiresome.
I’m using coding agents as a tool for software development and like to summarize complex matters. That doesn’t mean, that I like having AI “features” cramped into everything.
And also there are the valid ethical questions AI-apologists and enthusiasts tend to dream away.
You really don't understand how having an AI agent look through your wiki and pull up relevant links quickly is useful? Also, what higher chance of failure are you talking about. It's not like it's an integral part of the system. The whole Ollama component is entirely opt in. At least try to make some sense here.
In that particular case no.
My setup is searchable ZIM files on a pen drive. I open the file with a viewer on any PC or f-droid capable android. No servers involved, no handoff of data involved. More like a book as last resort.
Maps are openstreet maps downloads in osmand.
From my perspective there is no need to over engineer a basic tool.
For example, take a look at those TEOTWAWKI preppers. They’re hoarding books, ebooks, literature and knowledge in the most robust and reliable way. No need to add a tech stack that might or might not yield the proper answer and might or might not hide important facts in the summary.
Edit: https://feddit.org/comment/12160220 this fella has some great additions, too. E.g. Unnecessary resource hunger.
The advantage with the agents is that they can do much better fuzzy searching when you're not using exact terms, and they can correlate things for you. So, you can ask general questions and get useful information pulled out from your db. A searchable ZIM drive doesn't do that for you.
Too much AI, too much Internet dependence, need a complete distro including all data with single click install from USB stick and no downloading. IDK if Khan Academy even allows that.
It's obviously a different take on the "survival" data horde. Its obviously not intended as a wasteland survival kit, it seems intended for a scenario where the power is still on but universal internet access isn't a given.
Yeah, as a bunch of other people have said, the idea is good but the implementation is dubious. Khan Academy does allow downloading though: https://www.khanacademy.org/downloads
Implementation seems fine to me, I'm not sure what people think is dubious?
They say various things in the other comments. I agree with most of the criticisms.
What internet dependence, it's completely offline once you install it. Also, nobody is forcing you to use the LLM bundled with it. Stop perseverating.
self-contained, offline
While on one part i do like the all-ini-one ui and services, i feel as though it could have been done a little better without hosting a mini web-server just to use localhost on it.
Most if not all of the tools here are based on snapshots of online websites running in a browser, along with Docker ontop of it. While the intention is good and there are some neat ideas in here, why not just bundle native, offline FOSS programs that do the job already? for instance, cyberchef can be replaced with respective linux programs (eg base64, hexdump, grep, awk/sed, and gpg, just to name a few. graphical versions of these programs exist as well, so it's not like you need to use the terminal, it's just the most versatile environment for this type of stuff). No need for a webserver or anything.
However i will say, the offline wikipedia and maps are cool, unfortunately they're the only neat things in this project.
Now let's get to the point, an AI chatbot. What, does the dev think we have money to burn? Much less if SHTF and NVDIA RTX GPUs are scrapped for metal? (which they should be anyways). Now i know it's local, and that it most likely has data already trained on it so that it has the 100% guarantee of not huffing its own fumes and hallucinating, but compared to the absolute power usage that'll bring because of the sheer amount of resources it's hogging out trying to spit out an answer, a search engine could do just as good, and it won't hog up your GPU while at it. That's not even getting into the current ssd/gpu/ram situation right now. On its front page, its own recommended spec sheet says 32 gigs of ram. yeah that's a bit steep. 1TB SSD, i could kinda see why, but if i assume that most of the information is just text, you don't really need 1TB, but it is better safe than sorry. Still, that'll be pretty expensive if we're going by today's prices. When SHTF do you really think that most people are going to be rocking killer rigs with 8/16core CPUs, 32+ gigs of RAM and an RTX GPU? For the millionares and spoiled gamers who already have those? Sure, but for the masses? They'll mostly be using laptops with 4-6 cores, 8 gigs of ram, and a mid-range gpu if they're lucky, or integrated graphics.
Sure, you can say that having AI in it is somehow beneficial and tout how "everyone is using it", but don't get all pissy when your power bank runs out of juice at the worst time, let alone word gets out and your place gets raided and your 20-year-old 5090 is turned into scrap. All because you thought AI is good enough.
All in all it's a good premise, but it could be executed way better than just snapshotting websites, then slapping AI onto it and calling it a day.
I love the concept, i have even been working on something similar but, big buts…
Recommend ubuntu? While many are moving away from it.
Ai chat with ollama as a prominent feature? Controversy aside, this survival computer better packs some hardware, which may cost more precious possibly limited power.
Note taking app? Besides the intention to run it on ubuntu which i presume already includes something to work with markdown… any computer with a terminal can make notes as far as i know.
Hardware scoring and community leaderboard? wtaf
Things like offline wikipedia in kiwix are indeed pretty cool but in general the way this software describes itself feels sloppy and based more on vibes then anything though out.