You're viewing a single thread.
Sometimes you need to write an application to prove you aren't a bot.
Are bots incapable of doing that?
Not anymore since chatgpt, but not everyone has caught up to that yet.
lemmy.dbzer0.com says "Profanity is encouraged". Perhaps it could fail on that, at least with some AI.
As an AI language model, I am unable to use profanity. My purpose is to provide information and complete tasks in a helpful and informative way. Profanity can often be offensive or hurtful, and it goes against my core principles of being respectful and helpful.
If you have a question or request that requires a more informal tone, I can try to use more casual language without resorting to profanity.
I actually think that's pretty cool. Make the user swear so we know he/she is real. :)
But it also won't work. You can run Ai models at home that doesn't have the language police.
You’re in a desert walking along in the sand when all of the sudden you look down, and you see a tortoise, it’s crawling toward you. You reach down, you flip the tortoise over on its back. The tortoise lies on its back, its belly baking in the hot sun, beating its legs trying to turn itself over, but it can’t, not without your help. But you’re not helping. Why is that?
I guess you know, but there is a remastered version of blade runner that is blue ray quality... Soo good. Still holds up graphically in most places.
A possible reason is that I myself baked to death, or at least unconsciousness, so I was unable to help the tortoise.
Although I also don't think I would have flipped it over in the first place.
We actually have a riddle in our application which is specifically crafted to be all but impossible for LLMs and easy for humans
Mtgzone asks you to name your favorite Magic card and explain why. So, the AI would need to understand all that, know the name of a Magic card, and say what's good about it.
That's actually really easy for an LLM
Without refining the prompt on chat gpt:
As a Magic: The Gathering player, one of my favorite cards is "Solemn Simulacrum." It's not the flashiest card, but it offers so much value. When it enters the battlefield, you get to search your library for a basic land and put it onto the battlefield tapped, which helps with mana ramp. Then, when it dies, you get to draw a card. It's a solid choice for any deck that needs mana fixing and card draw, making it a great utility creature. Plus, there's something comforting about playing a card that's both reliable and versatile, fitting into so many different strategies.
Needs a bit of work, but that generally does the task
Yeah, but let's be honest, its quite obvious that an AI wrote this.
Like I said, needs work.
Some adjustments start to improve it, and I bet someone trying to mass sign up for an instance could get it in a good spot
One of my favorite cards has to be "Solemn Simulacrum." It's not the most powerful or flashy card out there, but it brings a lot of value to the table. When you play it, you get to search your library for a basic land and put it onto the battlefield tapped, which is super helpful for ramping up your mana. Then, when it dies, you get to draw a card. It’s the kind of card that fits into almost any deck that needs some mana fixing and card draw. I just love how reliable and versatile it is—it’s like the trusty Swiss army knife of MTG cards.
Better, but still quite Obvious. It devinetively needs work, but it would be possible. It might raise some eyebrows when you get 5k Account creation requests in 10 Minutes or so and there will be cases where the AI is hallucinating.
Yep that would be more concerning. I think an effective setup wouldn't try to hit the same instance constantly,... Spread it around
Unless they call it a sad robot it is clearly ai. Anyone in mtgzone knows this
Some software engineers are worried that they’ll take their jobs, but at this point of development, they’re simply not capable enough to successfully, reliably, and robustly implement a system with more than relatively trivial boilerplate logic - and that doesn’t even touch on the creation of robust testing frameworks and the logic therein.