ClioSport.net

This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • When you purchase through links on our site, we may earn an affiliate commission. Read more here.

ChatGPT (AI)

I've been preferring Copilot recently as seemingly infinite uploads... but I did finally meet the chat limit the other day,

Anyone done vibe coding? I haven't a clue to be honest but wouldn't mind giving it a go for a basic game.
 
I did my dashboard computer thing with Claude. 3 games, Bluetooth interface, settings menu etc took a few days and f**k tones of prompts but as I can’t code it was pretty fun.

Can play it here, works on desktop better.R129 RUN

IMG_9319.webp
IMG_9313.webp
 
Mr 800 threads just changed role, i bet his usage doesnt go down.
I can't help but think we may as well just use an agent in his place, given his clear dependency on it.
 
We have a PR firm write a report every year based on the hundreds of entries into the international business awards that we run. Its a lot of work and gets expensive as they have to read hundreds of case studies, pick up trends and then create the narrative for the report, draft, edit, make it look pretty. It also takes them around 2 months to complete.

I asked claude to do the same thing yesterday out of curiosity, it banged out a near identical report (if anything slightly better from the insights it picked up on) to the agency in about an hour. I know that the agency didn't use AI to pick out the trends it was done manually.

In the future we will probably be able to pull out all the trends from claude and just bounce it over to an agency to write it properly and design it into something nice looking saving us a lot of money on their day rate!
 
Human created content will start to rise in value. AI slop in the likes of films, music and games is already being ridiculed. There's a distinct irony in the fact that as AI becomes more flawless and able to make fewer mistakes - that the flaws themselves which humans create, become more desirable and thus more expensive to buy.

For now, even simple information has to be checked and rechecked. I was looking at some stats at the weekend after the BTCC season closer and ChatGPT contradicted itself.


1791961793367.webp
 
Chat GPT is a bit w**k at that sort of stuff TBH and tends to uncritically regurgitate whatever it pulls in from web scraping.

But the point still is a good one, validate everything as independently as you can.
 
Interested to know what model / subscription you were on for this?

Always worries me that I'm relying on it too much but I pretty much always use a fairly high model and it seems to do a decent job of checking its own working. At the minute that means Astra / Sol High on ChatGPT and Opus 5.5 High on Claude.
 
Back
Top