Started By
Message

Seems like Claude is the best AI.

Posted on 5/18/26 at 5:30 pm
Posted by T1gerNate
Member since Feb 2020
3347 posts
Posted on 5/18/26 at 5:30 pm
Feels like Anthropic’s Claude AI chatbot is just clearly better than chatGPT, Gemini, and Grok at this point.
Posted by WhiskeyThrottle
Weatherford Tx
Member since Nov 2017
7248 posts
Posted on 5/18/26 at 5:32 pm to
Claude Mythos is going to be wild.

I honestly hope it is never released. But the writing on the wall says it will and it's got some real capabilities that are some doozies.
Posted by Obtuse1
Westside Bodymore Yo
Member since Sep 2016
30496 posts
Posted on 5/18/26 at 5:37 pm to
Which is better depends on what you are using it for. None has stepped ahead to be the best at every task.
Posted by Decatur
Member since Mar 2007
32754 posts
Posted on 5/18/26 at 5:38 pm to
I have historically been very skeptical about AI but Claude is turning me into a believer.
Posted by Indefatigable
Member since Jan 2019
37325 posts
Posted on 5/18/26 at 5:40 pm to
Has it gotten better?

I’m a lawyer, and Claude still just completely makes things up more often than not every time I try and play with/test it. Just like the others.
This post was edited on 5/18/26 at 5:41 pm
Posted by CAD703X
Liberty Island
Member since Jul 2008
93336 posts
Posted on 5/18/26 at 5:41 pm to
quote:

Feels like Anthropic’s Claude AI chatbot is just clearly better than chatGPT, Gemini, and Grok at this point.
not convinced.

i have used both heavily; and i found chatgpt code (at least for my purposes) about equal to what claude outputs.

in the beginning i thought claude had the edge, but now i'm not as clear.

also this shite is constantly getting refined so what has the edge today, may not have the edge a week from now.

claude & chatgpt are going to be the 2 heavyweights for a long time but of course trillionaire elon has unlimited money to fuel grok so i don't see it going anywhere.
Posted by Lgrnwd
Member since Jan 2018
8738 posts
Posted on 5/18/26 at 5:43 pm to
Ok Mrs. Claude
Posted by WhiskeyThrottle
Weatherford Tx
Member since Nov 2017
7248 posts
Posted on 5/18/26 at 5:44 pm to
Gemini seems to do a good job at answering generic business related questions.

I have a CoPilot license for work that uses the Claude models for computing. The amount of code it can crank out is insane. I spent about 30 - 45 minutes typing out some steps I take to do something on a regular basis. Maybe an hour or two a week of work, but it's several steps and if I were to have coded it, it would have taken a week or two. Son of a bitch knocked it out damn near perfectly the first time in about 15 minutes. Absolutely unreal how much AI has progressed from 4 or 5 months ago. Developers who are hesitant to use it are the ones that will get left in the dust. Developers who understand the business they're working for and can prompt AI effectively are the ones who will get ahead. Developers who understand how to create markdown instruction files with high levels of instruction, and prompt AI and understand how the business works and the direction they're headed are going to get paid big time. I was hesitant to start using AI until about 3 months ago when someone demonstrated what it can do. Downloading Powershell 7 and running the CoPilot CLI has been a huge help with some of the stuff I've inherited in the last 3 months.
Posted by BoogaBear
Member since Jul 2013
7345 posts
Posted on 5/18/26 at 6:06 pm to
There are caveats to that as well, Amazon Q, kiro, and cortex code all use Claude and I find there are variations to them.
Posted by Volvagia
Fort Worth
Member since Mar 2006
53467 posts
Posted on 5/18/26 at 6:07 pm to
The problem with these cross comparisons is no one ever says what mode they are using them in.


Aka they are too stupid to even evaluate output. Garbage in, garbage out.


Example: standard chatGPT and GPT in research mode are light years apart.

In the latter it spent nearly an hour crunching on it (after asking for follow up questions for clarity in initial request) really digging deep on the topic. And I’ve always been satisfied professionally. Still review it though.


Standard ChatGPT as often as not is just trawling Reddit posts.
Posted by McLemore
Member since Dec 2003
35329 posts
Posted on 5/18/26 at 6:09 pm to
quote:

Has it gotten better? I’m a lawyer, and Claude still just completely makes things up more often than not every time I try and play with/test it. Just like the others.


I use co-work projects in conjunction with my go-by database and WestLaw (whose AI is laughably bad).

Claude is incredibly good when properly prompted, sources, and checked.

Eta: and it’s way better now than 3 months ago. Way.

This post was edited on 5/18/26 at 6:12 pm
Posted by CAD703X
Liberty Island
Member since Jul 2008
93336 posts
Posted on 5/18/26 at 6:09 pm to
quote:

have a CoPilot license for work that uses the Claude models for computing. The amount of code it can crank out is insane. I spent about 30 - 45 minutes typing out some steps I take to do something on a regular basis. Maybe an hour or two a week of work, but it's several steps and if I were to have coded it, it would have taken a week or two. Son of a bitch knocked it out damn near perfectly the first time in about 15 minutes. Absolutely unreal how much AI has progressed from 4 or 5 months ago. Developers who are hesitant to use it are the ones that will get left in the dust. Developers who understand the business they're working for and can prompt AI effectively are the ones who will get ahead. Developers who understand how to create markdown instruction files with high levels of instruction, and prompt AI and understand how the business works and the direction they're headed are going to get paid big time. I was hesitant to start using AI until about 3 months ago when someone demonstrated what it can do. Downloading Powershell 7 and running the CoPilot CLI has been a huge help with some of the stuff I've inherited in the last 3 months.


i created a MONSTER Home Assistant setup all using AI. 15+ dashboards, warnings, alerts, color changes; live map view of my robot vac as its cleaning; rachio irrigation tied into weather alerts, fitbit dash with celebration animations after reaching certain daily milestones, security camera setup; people tracking, auto-door-locking if someone is appraoching the house, smoke detector integration, garage door, weather radar loop triggered by NWS alerts coming on full-screen, special 'morning' and 'evening' dashboards..

and forget about the eye candy above..the way it hardened the setup on both my mini-pc 'display brain' and my HA PC server; addressed memory leaks when the dashboards were running too long; helped set up samba shares, virtualbox, eufy proxy for door locks, creating best-practice daily backups & 7 day overwrites, auto-off-site backup creation; fixing random crashes the RIGHT way instead of just ignoring them and hoping they don't happen again...


yes i had to know enough to ask it to do those things; but this is fricking killer. these are things i would have 'wanted' to do before but too lazy or lack the time to figure it out...

so its more than just the coding; its the entirity of the ecosystem it creates that turns this from a simple hack into a robust platform you can be proud of and won't crash and burn.
This post was edited on 5/18/26 at 6:11 pm
Posted by Bestbank Tiger
Premium Member
Member since Jan 2005
80891 posts
Posted on 5/18/26 at 6:16 pm to
I haven't gotten good results yet.

I asked it for the closing prices of a series of stock tickers on a certain date and when I checked it against the actual data it got several of them wrong.

I asked it to put Coach O's face on a picture and the guy they used looked nothing like O.
Posted by Bison
Truth or Consequences
Member since Dec 2016
1313 posts
Posted on 5/18/26 at 6:22 pm to
I don’t get the obsession with AI.

Y’all realize that’s just making us dumber as a society right?

frick AI.
Posted by Decatur
Member since Mar 2007
32754 posts
Posted on 5/18/26 at 6:34 pm to
quote:

I asked it for the closing prices of a series of stock tickers on a certain date and when I checked it against the actual data it got several of them wrong.


It’s not magic.
Posted by BeerMoney
Baton Rouge
Member since Jul 2012
8923 posts
Posted on 5/18/26 at 6:34 pm to
I pay for the pro version. I use it for so many things. It’s amazing
Posted by Bestbank Tiger
Premium Member
Member since Jan 2005
80891 posts
Posted on 5/18/26 at 6:52 pm to
quote:

It’s not magic.


It should be able to retrieve information.

Main reason to ask for past market information is it can look it up faster than you can.
Posted by SallysHuman
Lady Palmetto Bug
Member since Jan 2025
21745 posts
Posted on 5/18/26 at 6:56 pm to
quote:

I don’t get the obsession with AI.


It's useful. It's fun. Pretty simple, really.
Posted by Decatur
Member since Mar 2007
32754 posts
Posted on 5/18/26 at 7:08 pm to
My bad I misread…presumed you were asking it to predict a future price. For best results you can use someone’s API. Without that results look really spotty for that query.
Posted by HoustonGumbeauxGuy
Member since Jul 2011
33509 posts
Posted on 5/18/26 at 7:12 pm to
My company is doing a pretty widespread trial run with Claude, we use copilot for some shite which is helpful. I’ve been using ChatGPT for a long time so it knows my tone the history of my asks, etc. so it’s answers are magnificent, but I keep hearing all this cool shite about Claude.

first pageprev pagePage 1 of 2Next 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