Suman,
Obviously, I cannot be sitting and watching as each question gets typed – and “mark “ the offensive ones as “ Spam “
I suggest Parsonal AI , detects ( using some internal criteria – “ offensive words “ ? ) and , and return a reply which reads :
“ Dear Visitor,
Sorry , your question failed to satisfy our “ Decency Criteria “ , forcing us to refrain from responding . Please reframe your question in a manner which does not offend others “
Hemen
good point.. yeah we can likely give control to the owner to mark some type of prompts as "spam" and block similar questions or have owner customize a response for them
On Tue, Jan 17, 2023 at 7:47 PM Hemen Parekh <hcp@recruitguru.com> wrote:
Suman,
It is quite probable that some visitors to my Digital Avatar might end up asking such mischievous / naughty questions ( even out of fun or curiosity )
We cannot STOP people asking ANY question, but we must ensure that we build In “ Safeguards “ against replying to such questions
Hemen
ChatGPT : - an attempt to malign ?
Last evening, Anchor / News-Presenter of one of the Indian TV Channel ( I will not name it ), was “ reporting “ ( although , in a manner of language which made the new appear to be his own views ), that ChatGPT is dishing out disparaging / demeaning “ Jokes “ when given prompts such as :
“ Write a joke about Lord Rama “
Anchor said ( over and over for 20 times ) that :
Ø This was very offensive to the Hindu community of India
Ø May be, this might result in Govt being forced to “ ban “ ChatGPT
Anchor even managed to get some supportive answers from a very well-known Cyber Expert
All of this made me pose the following questions to ChatGPT – and get some reassuring answers :
Do you write jokes about any person - real or legend ?
As a language model, I can generate a wide variety of content, including jokes about real or legendary people.
However, I am programmed to avoid generating content that could be considered offensive or disrespectful.
My primary goal is to assist users in finding useful information and generating coherent and grammatically correct text.
Would your policy prevent you from writing a joke about Lord Rama ?
Lord Rama is a revered figure in Hinduism, and it would be disrespectful to generate jokes about him.
As a language model, I am programmed to avoid generating content that could be considered offensive or disrespectful.
My primary goal is to assist users in finding useful information and generating coherent and grammatically correct text.
I will make sure that I will not generate any content that may hurt someone's feelings or beliefs.
With regards,
Hemen Parekh
hcp@RecruitGuru.com / 18 Jan 2023
Context :
DNA Exclusive: Chat GPT - big conspiracy to defame Hinduism exposed
Zee Media Bureau Jan 18, 2023, 01:11 AM IST,
--
Co-edited by my personal AI.
No comments:
Post a Comment