Something somethING ICE in bed with Telcos to deport foreigners buying burner phones?

Please let it be stupidly implemented such that I can convince the AI to pay me to fly Delta. (IMHO, how are they even a big airline player? I give it they are a step above Spirit but that’s all they got)

But for the lulz, what was its response and why was it more than 1 word? (I have yet to see a non single purpose driven Instruct AI actually adhere to the single word directive on a complex topic.)

When the Distro starts talking about enterprise features during the installation process (looking at you canonical)

Last I heard, they use Llama-2 since its the only one approved for GOV work.

But its probably Grok because they don’t seem keen on approvals

Oh man deer are F R E A K Y when it comes to disease carriers They got the mad cow prions, the chronic wasting prions, natural reservoir for COVID, hemorrhagic diseases, west nile…. They are odd animals.

My theory? Its the abstract that trips it up. I don’t know much about neural networks or RAG searching, but the abstract is information dense in a short amount of space. My guess, an LLM scans through a document, finds areas of the document that has the largest quantity of information relevant to the prompt and then branches out until its confident enough in answering the question. While great for most written text, I think the LLM starts with the abstract, gets way too much confidence that it knows enough and then begins spouting bullshit. In my opinion it explains why it creates fictitious citations. Probably sees a citation in an abstract and then conjures up a title from the information in the abstract.

My only evidence is that I try using LLMs to find information quickly in very dense datasheets for electronic components and most times it will rattle off information found in the component overview section rather than telling me about information found in the application, or electronic characteristics section.

Disclaimer: If you want to explore window managers then go ham! Linux is all about exploration.

Now, If you think the grass might be greener on a different desktop manager then stick with gnome. By no means am I saying Gnome is the best, but its more of a situation where it will devolve into the quirks you know vs the quirks you don’t know situation.

Personal Antidote, I started with Gnome and used Gnome for years. Got curious and started jumping around I tried KDE, I3W, XFCE, Pure X, Etc. There were things I liked about each one of them but the quirks of each deviating from my expectations coming from gnome was too much and I ended up sticking with gnome.

That being said, out of necessity due to system constraints I run XFCE when I need a light weight DE. A close second in that realm is LXDE But I don’t like its default aesthetic nor do I feel like customizing it since I do most of my computing in a terminal.

The embedded IoT crowd would like to refute your claim that there are no operating systems that you can install and forget.

The collective would like to stress that any operating system can be installed and forgotten. Please note, that usefulness and security may be impacted.

/s

Also, to be technical there is CollapseOS which is an install once and forget sort of thing.

My first new computer was an Acer Aspire One netbook with Windows 7 starters. I quickly realized what “starter” meant and discovered Ubuntu 10.10 Netbook Remix. The rest is history.

Not sure, currently have 8 nodes and 40 apps running

Use tailscale for host nodes, use tailscale docker container in a compose stack with an app that you sidecar to. That way that app is on your tailnet as if it is its own computer. Use tailscale serve for reverse proxying support of the apps. Then, setup a vps node (I use linodes $5 node) with tailscale and configure that to be your DMZ into your tailnet.

For DMZ, use Caddy, UFW, and fail2ban. Also take advantage of ACLs in the Tailscale admin console to only have the VPS able to route traffic to specific apps you want to expose. My current project is to work in Authelia into this setup so a user logs into one exposed app and is able to traverse to other exposed apps through header / token authentication.

Oh also, segment the tailnet using different authentication keys. Each host node should have its own key, all the apps on a host node should have a shared key, and all public facing clients should have a common shared key. That way in case of compromise you can revoke the affected keys without bringing down your network.

When you end up having a mini homelab look into komo.do for container orchestration over the overkill options like kubernetes or portainer

In windows defense (no means sticking up for them now) It was a pretty unobtrusive OS in Windows 7 and arguably in Windows 8 (but don’t get me started with the UI/UX choices). Windows 10 was decent and for the first year or two felt good running it. But after that yikes…… Then windows 11 comes to the scene and I lost the plot. Looking forward to October though when people throw out their 7th Gen Processor rigs. I got no issues rocking an I7-6700K that is not AI ready

.exe to .sh low key turn all windows machines to Linux machines