5 ms·
It's so lazy and I (probably most here) can spot the AI writing style instantly. They continue to give nonsense answers and waste time of open source maintainer
by hypeatei 3y ago
It's so lazy and I (probably most here) can spot the AI writing style instantly. They continue to give nonsense answers and waste time of open source maintainers.
- the_af 3y agoSomething about the excessive politeness and constant apologizing and clarifying gives it away, right? Just wait until people use LLMs with rude, impatient styles and broken English. Maybe an LLM that goes into flamewar tangents, maybe makes unrelated racist remarks, etc. Then we will have truly reached terminal confusion.
- dexwiz 3y agoWhat makes you think there aren't people out there running models without the corporate safety and alignment layer? For the average person, this is a big ask, but not for a corporate or nation state actor.
- fnordpiglet 3y agoIt’s not that big a task as all that. There are a lot of unaligned models available, and user interfaces that aren’t that hard to use. https://github.com/henk717/KoboldAI https://github.com/henk717/KoboldAI
- dexwiz 3y agoAnything that requires reading a github page is a big ask for an "average person." Maybe not for you or me, but we are in a bubble here.
- fnordpiglet 3y agoBut we also aren’t a megacorp or a nation state, and while as a percentage of the world we aren’t a huge number, as an absolute number there are a lot of people capable of doing this.
- hypeatei 3y agoYes it's like corporate speak and very long winded. Also the way it outputs headings and lists is very identifiable.
- CoastalCoder 3y ago> Something about the excessive politeness and constant apologizing and clarifying gives it away, right? That's part of it for sure. Somehow it manages to be obsequious, patronizing, corporate, disingenuous, corporate, useless, and passive-aggressive, all at the same time.