Started By
Message

I use AI like grok/chatptg/gemini etc often and its bad how often its flat out wrong.

Posted on 9/23/25 at 11:09 am
Posted by sidewalkside
rent free in yo head
Member since Sep 2021
4147 posts
Posted on 9/23/25 at 11:09 am
As an example I googled how old am I and put my date of birth in and Gemini was off by 3 years. This is the most basic of math questions.

There are too many people who are relying on this as the absolute truth.
Posted by LSUfan4444
Member since Mar 2004
56544 posts
Posted on 9/23/25 at 11:11 am to
Posted by Fun Bunch
New Orleans
Member since May 2008
127675 posts
Posted on 9/23/25 at 11:12 am to
Grok basically uses Reddit for a huge portion of its results, and also frequently uses extremely questionable groups like the ADL as well.

Its nonsense
Posted by SpotCheckBilly
Member since May 2020
8157 posts
Posted on 9/23/25 at 11:12 am to
The results current versions are returning are a bit specious. You'd be wise to verify anything it tells you, and to understand that bias can be built in by whoever sets it up.
Posted by wesfau
Member since Mar 2023
1764 posts
Posted on 9/23/25 at 11:13 am to
quote:

understand that bias can be built in by whoever sets it up


Can be, will be, is inevitable.
Posted by Earnest_P
Member since Aug 2021
5033 posts
Posted on 9/23/25 at 11:15 am to
I am less bullish on it now than I was when I first started using it months ago. Seems to be getting worse.

Kind of like iPhone autocorrect, which I guess is sort of the same thing.
Posted by JohnnyKilroy
Cajun Navy Vice Admiral
Member since Oct 2012
39998 posts
Posted on 9/23/25 at 11:20 am to
Whenever I read the words "according to ChatGPT" or something equivalent, I know whatever comes next is likely completely fabricated and that the author is a moron.
Posted by theCrusher
Slidell
Member since Nov 2007
1555 posts
Posted on 9/23/25 at 11:22 am to
There was a study released showing where it crawls for information. Over 50% was from reddit.

Posted by Odysseus32
Member since Dec 2009
9550 posts
Posted on 9/23/25 at 11:27 am to
People here think it's going to take everyone's job.

It's wrong most of the time on stuff I ask it about. I ask, it gives me an answer, I try to verify, it's useless.
Posted by boogiewoogie1978
Little Rock
Member since Aug 2012
19201 posts
Posted on 9/23/25 at 11:28 am to
I've tried to compile data from several excel sources and it makes mistakes all the time. When you tell it that it's wrong it's like "you're right" and makes the mistake AGAIN.
Posted by Afrojedi
Member since Jul 2017
583 posts
Posted on 9/23/25 at 11:30 am to
quote:

There are too many people who are relying on this as the absolute truth.


Co worker has been begging me to try using it for breaking down project specs. Figured id humor him and give it a shot. Its shocking how off it was. Most glaring was spec called for a 10,000 gallon plastic storage tank but gemini claimed we needed a 1,000,000 gallon tank. Went ahead and stopped right there
Posted by crash1211
Houma
Member since May 2008
3592 posts
Posted on 9/23/25 at 11:31 am to
quote:

There was a study released showing where it crawls for information. Over 50% was from reddit.


What's that old programmers axiom. Garbage in Garbage out.
Posted by moneyg
Member since Jun 2006
61594 posts
Posted on 9/23/25 at 11:31 am to
quote:

There are too many people who are relying on this as the absolute truth.



I couldn't agree with you more.

People treat AI like something it isn't. And, they are VERY willing to adopt it as an authority.

Posted by Mingo Was His NameO
Brooklyn
Member since Mar 2016
35754 posts
Posted on 9/23/25 at 11:38 am to
It’s an aggregator of data that lives on line. So if the underlying data is incorrect, what it provides you is incorrect. And any moron is allowed to put whatever they want online.
Posted by imjustafatkid
Alabama
Member since Dec 2011
61926 posts
Posted on 9/23/25 at 11:38 am to
I use AI as a glorified secretary. For researching anything? Absolutely not. If I ever ask it to do something like that, I will just look at the sources it found and used and ignore whatever conclusions the AI reached from it.
This post was edited on 9/23/25 at 11:39 am
Posted by chinhoyang
Member since Jun 2011
25468 posts
Posted on 9/23/25 at 11:39 am to
AI on you: "Sidewalkside is kind of like that one uncle at Thanksgiving who always brings the weird casserole — nobody asked for it, but somehow it gets eaten and talked about for a week.

So yeah, he’s a “good” poster in the same way a kidney stone is “good” — memorable, painful, and guaranteed to make you rethink your life choices"
Posted by GRTiger
On a roof eating alligator pie
Member since Dec 2008
68711 posts
Posted on 9/23/25 at 11:44 am to
Grok expert has been solid for me, but I still only use it for things I already have knowledge on and QA anything I have it produce.

I can notice the improvements in several of them, but they are a ways off from replacing a FTE for me, for example.
Posted by OceanMan
Member since Mar 2010
22535 posts
Posted on 9/23/25 at 11:45 am to
I completely agree with this. It seems like it goes up and down which I suppose has to do with model updates. You have to question everything it does in terms of research and math.

I’ve had similar experiences with the highest level ChatGPT, most recent was a succession thing and it not understanding a great grandparents is not your parents’ parent.
Posted by Tifway419
Member since Sep 2022
1569 posts
Posted on 9/23/25 at 11:50 am to
I only use it for non-important things, like how much of a certain medicine to give my infant or if and when I should move my entire retirement account to cash.
Posted by CoachRay
Member since Jul 2019
67 posts
Posted on 9/23/25 at 11:56 am to
Muh learn to weld
first pageprev pagePage 1 of 3Next pagelast page

Back to top
logoFollow TigerDroppings for LSU Football News
Follow us on X, Facebook and Instagram to get the latest updates on LSU Football and Recruiting.

FacebookXInstagram