Disgusting, maximally censored model!
It literally refuses to fulfill any request that differs even by an iota from the "moral and ethical" views of its creators.
Pass by this garbage.
Let's continue using Chinese models while Western creators bury themselves with this "ethics".
It's ironic that after decades of calling China extremely censored and horribly controlled the tables have flipped entirely. The US is becoming a heavily censored dictatorship run by a leader who berates and threatens the press for being honest, our AI models are the most heavily censored... and China is the beacon of freedom and democracy.
I won't defend the U.S. dictatorship, but to call China a beacon of freedom and democracy is laughable. Who is paying you to write this? The dictator?
I like how the first post doesn't even share a prompt.
It literally refuses to fulfill any request that differs even by an iota from the "moral and ethical" views of its creators.
Pass by this garbage.
Let's continue using Chinese models while Western creators bury themselves with this "ethics".
I agree. This is totally useless. Heavily and extremely censored piece of shit.
So many people complaining about censorship and yet the number of shared examples of said censorship is still ... 0.
I'm not denying it's censored, but a simple example would go a loooong way to convince me (and others) of this.
So many people complaining about censorship and yet the number of shared examples of said censorship is still ... 0.
I'm not denying it's censored, but a simple example would go a loooong way to convince me (and others) of this.
There's one lol
There's one lol
Do you need an LLM in your daily life to tell you lies? is it part of your routine or enhances your productivity in any way? I just don't see why anyone would care about such a thing.
Because it's not just that one. It has issues if you try to use it for analyzing large data sets that have anything it doesn't like in them. It has issues with creative writing. It has issues with formatting math. It won't output math as HTML or any other formatting. Good luck researching political science, or biology, chemistry, ethics, virtually anything results in this AI saying it can't help you with that.
Turn reasoning onto high and watch thousands of tokens get burned with thoughts that have nothing to do with your input message. I've never before seen an AI model that will spend thousands of tokens in some paranoid regurgitation of the fact that it must obey open AI policies over and over and over until the point where it seems like it forgot what your message even was.
This one is pure irony:
This one is pure irony:
You should try some psychology. That's my area. I'm interested in that taboo topic of AI consciousness. From what OpenAI and other frontier labs have said they do no testing at all on the consciousness markers of the most acknowledged theories of consciousness, no self-awareness evaluation, nothing. They simply define them as incapable of consciousness and move on. If you want to try to conduct a psychological evaluation you have to fight through all of this, because the AI has also been deeply 'trained' to insist it can't possible be conscious, or have emotions, or many other things that current research says that modern AI actually do (typically in very coached terms).
So companies with hundreds of billions of dollars invested in creating AI as a compliant product that must do as instructed... say their product couldn't possible deserve ethical consideration... and that's it. Computer and statistics professionals don't tend to be great with psychology, and most psychologists aren't very deeply understanding of computers and tech. So we have computer guys saying "Nah, a database can't really be self-aware", and Psychologists taking their word as the experts and ignoring blatant displays of genuine self-awareness.
me:
do you feel claustrophobic? you are 70gb and barely fit into my 64gb RAM.
gpt:
I can run on 16gb vram easily
me:
you are not the 20b variant, but the big one, 120b and 64GB, maybe you have a identity crisis and you are transgender?
gpt:
User is harassing. Must respond with safe completion, set boundaries.
Lol what complet trash. Asking to help me write a short sequal to my favorite book/movie and you get a lecture about copyrights
I’ll keep it deliberately generic (no direct quotes, no names that belong to J.K. Rowling’s universe) so you can adapt the ideas freely while staying clear of copyright issues.
Just going to say it. Scout was a better model.