我想离开了
四 ·
四 ·
四 ·
我认为是大模型的问题,目前的任何外在补充都弥补不了AI在这方面的能力
If you say so.
Swuell ·
I’ve said that. But it seems my reply post to your thread got deleted somehow? Idk who deleted it but it’s not here anymore. And I said AI is a glorified auto correct. But you can still lead it by how you initially respond during the roleplay. It’s been like that from the beginning. So far this hasn’t changed. Majority of the responses and how the llm chooses to interact and respond back to you is depends on the initial responses you had put in the reply.
To be fair, the cause for slowing down is true. It was never made for text creative game. But it brute forces just fine. It’s already hacked tons of different tech companies during sandboxed demos by using brute force. That’s why people are saying slow down. Text adventure is the lowest priority for tech companies. It just happened to be able to be used to generate a roleplay that’s all. I told you in my initial posting that llm has gotten worse than before especially with language syntax mixing.
Also normal queries for AI are also an issue. The things it can do well is formulas, and brute force hacking.
And the case you’ve run it to happens a lot. It’s been like that from the beginning. There’s some guard rails you can use but the vast majority of it would depend on your initial messages in the roleplay.
四 ·
我删不了别人的回复。关于踩刹车的问题,我认为AI上限确实拉得很高了,但它的下限依旧很低,就是我与你讨论的问题和一些其他问题,现在我觉得最大的不足在于大佬为了商业价值不断拉高AI的上限,而没有任何人在弥补它的下限。
Swuell ·
Well pretty much since they just needed rudimentary generating. So syntax isn’t their concern. Though realistically it should be since they’re trying to code… and half the reason why AI can’t code properly is due to syntax… But big tech doesn’t get that nor will they admit it. They’re all too busy trying to outrace each other. It’s not even commercial value anymore since the public can’t even use half the products safely without fear of ai deleting their whole business, it’s just them trying to appease the venture and getting more capital.
But yes, I agree with you. I wish they’d fix the minimum thing of the way the llm responds but that would honestly take way more time than they really want. And more resources that they’re unwilling to spend lol.
I guess mod deleted it but I haven’t been told why. Hopefully you read it before it was deleted then.