The AI of Google resolv a lot of computers problems

Logicien

Logicien

I was not able to create a working bridge in /etc/rc.conf after many tries and it do it for me:
# Bridge0 Tap0
cloned_interfaces="bridge0 tap0"
ifconfig_em0_name="eth0"
ifconfig_igb0_name="eth1"
ifconfig_igb1_name="eth2"
ifconfig_eth1="up"
ifconfig_eth2="up"
ifconfig_bridge0="inet 192.168.7.1 netmask 255.255.255.0 addm eth1 addm eth2 up"

I was not able to make the things work with aufofs and him too !! But he redirect me to smbnetfs who work pretty well. He help me and verify my configurations files /boot/loader.conf and /etc/rc.conf even /etc/pf.conf. I am very weak with firewall rules. I think that the AI make things easier with Unix like systems and therefore can bring more people to use them.
 
There's lots of strong opinions for and against AI. As SirDice once pointed out, it's more or less like a search engine on steroids, (I hope I'm not twisting his meaning) rather than true artificial intelligence. For myself it once helped me solve a configuration problem that I couldn't figure out, but then I found myself getting lazy and just going to claude.ai rather than doing my own web searching.

The other bad thing about it is that it's sometimes completely wrong, so you have to be careful. In my (limited) experience, the sort of problem you mention, how to configure a bridge, is the sort of thing that it's probably good at doing.
 
I want to emphasize what scottro said. I, too, have started using it more often--and often accidentally--as a Google search winds up actually be answered by Gemini. I'd say one out of every three questions answered are either sloppy, wordy or flat out wrong but my questions are typically very technical.

If you look at Gemini, it gives its references for its answers. That is, it shows that it got its answers from other sources and did NOT think things up on its own. Those references can be from unofficial or untrusted sources such as reddit. I just woke up so I'm not sure if chatgpt also gives its references but I don't think so.
 
it's more or less like a search engine on steroids
Click to expand...

Thats right, the issue is google search is so bad these days
most of the time the first couple of pages are just spam posts

you dont get many people with blogs these days
so you end up with short posts on things like medium without any actual info
 
Like @drhowardfine (heh, we're referencing each other back and forth here), I do find myself using it more (too?) often. But, I remember vaguely, talking to my niece about it last Thanksgiving and I don't think my opinion's changed. At the time I said, for simple stuff it's good but if I ask some sort of tech question, like how to do something with a FreeBSD jail, the answer will often be no better than a startpage (which I use instead of google) search. Also as the good doctor said (hrrm, if we get him angry we can mad doctor), it's often an accidental thing. Really, in all this time I think there's only one thing that I just couldn't find with a web search, (how to password protect a page with nginx--problem was I didn't realize it had to be the https section, not general http section). Just about every other time I've used it, was because I was lazy or grouchy--e.g., my .maildrop filter wasn't working and I didn't feel like finding out what I'd typoed, so asked What's wrong withi this file?

Conversely, (sort of), I know a guy who's been railing against it and just recently wrote me that as his company has some sort of account with it, (ChatGPT, maybe?) he's decided to start using it for some stuff. (Stuff with his life, don't know if it will be successful or not).
 
Can you imagine 55% of citations are FAKE and yet people are like DERP give me more..............

This in major scientific publications.

People should be executed for FAKING science. At least have some public humiliation trail showing their stupidity and waste of funds.

Q&A from the slop trenches – GeoSpatial ML

We reviewed 22 ML conference submissions this summer. Fifteen had fabricated citations, hallucinated authors, or clear LLM slop, so we complain about that a bit, and release the paper references audit we now run.
geospatialml.com geospatialml.com
 
a program that look in databases with a "learning search" trained method...
but how explain it in what my mind thinks?

1.gif
 
Sometimes I slip up and instead of Startpage.com I type in Google.com and on LibreWolf it proceeds to ask me if I am a robot and sends me through CAPTCHA after which I grow bored and go back to StartPage.

Thanks mister google there is nothing more insulting than that. CAPTCHA because I have secure browser settings.
 
Logicien said:
I was not able to create a working bridge in /etc/rc.conf after many tries
Click to expand...
That's where I'd just not be creating a bridge before introducing stuff I wasn't able to figure out sooner from man pages/etc :p

I scroll-past Google/DDG's initial AI answer (don't want the habit of trusting it); even if a glance looks like the answer I still look at other pages to be sure.
Phishfry said:
Sometimes I slip up and instead of Startpage.com I type in Google.com and on LibreWolf it proceeds to ask me if I am a robot and sends me through CAPTCHA after which I grow bored and go back to StartPage.
Click to expand...
I've seen captcha with DDG and StartPage too (I rotate around em whenever they want to be captcha-happy out of nowhere)
 
I have become sceptic prudent with the recommandations of the AI of Google concerning the wireless cards. It recommand me some models that can do what I want on FreeBSD and after I have buy one it tell me that I cannot do what I want with this card.

I have a Kamrui Mini PC who have stop to respond, no light nothing. It tell me to buy a new power adapter but, I think it's life have rich the end. I cannot test the actual charger to know if it work so, I wait to do something.
 
AI, even the free one, is very helpful with prototyping.
It gives you basic syntax and options fast, even the general logic/schema. No need to sift through google searches (mediocre general reddit posts mostly) and copy/pasting code from stackoverflow like in the old days.
Then you can quickly test the proof of concept, for sometimes it reads some obscure post on a forum, only suggesting some feature but it "thinks" the feature is there and even hallucinates syntax for it.
For production I would recommend building it block by block, segment by segment. Actually, exactly with accordance with Unix philosophy.
Some coding harness (Pi, opencode] can probably manage this on it's own. Plus being documentation on the fly. But with the risk of losing perspective - crucial if buggy.

If I embark on some "bigger" project, I usually search for Video proof. Some guy showing it working on youtube. In my experience, there is a lot of brainiacs who suggest to you solutions which should work in theory from behind the desk but in reality you run in to million nuances, sometimes fatal i.e. manufacturer did not implement a feature. It should work but it just doesn't.
 
You must log in or register to reply here.
Back
Top