3 ms·

There was a paper a while ago, where they found a strategy for GPT beating humans on the Turing test 2/3rds of the time. They just prompted it to talk the way

by andai 3mo ago

There was a paper a while ago, where they found a strategy for GPT beating humans on the Turing test 2/3rds of the time. They just prompted it to talk the way a 13 year old girl texts. (Also they prompted it to pretend not to know anything about anything.)

9dev 3mo ago
Most of the LLM tells we usually recognise are due to their heavy attunement to role playing a "helpful assistant" with a knack for correct typography. If you used a smaller corpus, followed a different kind of fine tuning routine with rather normal conversation patterns, and wrote a different base prompt, I don’t see why it shouldn’t be possible to come up with a very particular personality, so to speak.
DharmaPolice 3mo ago
I feel like pretending to be a child is cheating somewhat. Might as well print random key presses and claim it's a toddler.
- echoangle 3mo ago
  [dead]
Someone 3mo ago
> where they found a strategy for GPT beating humans on the Turing test 2/3rds of the time. IIRC, PARRY https://simulation-and-society.org/parry/ https://simulation-and-society.org/parry/ got there in 1972 by pretending to be a paranoid schizophrenic person
- andai 3mo ago
  >while ELIZA has a comforting effect that many crave, PARRY often actively makes users uncomfortable and afraid PARRY is my spirit animal. > In empirical evaluations conducted during the 1970s, experienced psychiatrists proved unable to reliably distinguish PARRY's responses from those of actual patients diagnosed with paranoid disorders, achieving only 48 percent accuracy, a figure statistically indistinguishable from random guessing. https://ai.miraheze.org/wiki/PARRY https://ai.miraheze.org/wiki/PARRY
- andai 3mo ago
  Also I found the source code! https://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas/classics/parry/ https://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas...
sire-vc 3mo ago
https://xkcd.com/2635/ https://xkcd.com/2635/

by andai 3mo ago

9dev 3mo ago

Most of the LLM tells we usually recognise are due to their heavy attunement to role playing a "helpful assistant" with a knack for correct typography. If you used a smaller corpus, followed a different kind of fine tuning routine with rather normal conversation patterns, and wrote a different base prompt, I don’t see why it shouldn’t be possible to come up with a very particular personality, so to speak.

DharmaPolice 3mo ago

I feel like pretending to be a child is cheating somewhat. Might as well print random key presses and claim it's a toddler.

echoangle 3mo ago
[dead]

Someone 3mo ago

> where they found a strategy for GPT beating humans on the Turing test 2/3rds of the time. IIRC, PARRY https://simulation-and-society.org/parry/ https://simulation-and-society.org/parry/ got there in 1972 by pretending to be a paranoid schizophrenic person

andai 3mo ago
>while ELIZA has a comforting effect that many crave, PARRY often actively makes users uncomfortable and afraid PARRY is my spirit animal. > In empirical evaluations conducted during the 1970s, experienced psychiatrists proved unable to reliably distinguish PARRY's responses from those of actual patients diagnosed with paranoid disorders, achieving only 48 percent accuracy, a figure statistically indistinguishable from random guessing. https://ai.miraheze.org/wiki/PARRY https://ai.miraheze.org/wiki/PARRY
andai 3mo ago
Also I found the source code! https://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas/classics/parry/ https://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas...

sire-vc 3mo ago

https://xkcd.com/2635/ https://xkcd.com/2635/