OpenAI is working on eliminating all the problems that make AI represent
human beings! If ChatGPT can't make any social mistake, no bias, no lies,
no: "Hallucinations, Harmful content, Harms of representation, allocation,
and quality of service , Disinformation and influence operations,
Proliferation of conventional and unconventional weapons, Privacy,
Cybersecurity, Potential for risky emergent behaviors,
Interactions with Other Systems, Economic impacts, Acceleration,
Overreliance " (see URL Below) then who is being represented by
this AI? Aliens? Not only it's no longer representing current
iteration of human beings but it will be boring like hell!
Assume a strong country A is opressing a smaller country B.
Country B is going to refer to country A as an invider. Country
A is going refer freedom fighters of country B as terrorists.
How is ChatGPT going to represent this? It seems that country A
views would have more media power thus its "Truth" would be
more represented on the Internet!
Besides, OpenAI is spending lots of resources to censure the
"True Humanity " in a democratic world!
I suggest to OpenAI to be fair and let us see 3 versions ChatGPT reponses:
A)Unfiltered responses
B)Sanitized from Human harming behaviors
C)The "Next Humanity" Version in 500 years which seems to be
what OpenAI is trying to create with their censure. This is
the version when humanity will be One Nation with beings
fighting for Cosmic survival beyond the Solar System!
Not only, options A, B are more represenative of humanity today,
but they are also useful to society because they show our shortcomings
when compared to option C. They may also be used by leaders and policy
makers to see the truth about the situations of their citizenry and
may be design policies to correct the human trajectories!
These are my early "hallucinations" I'd love to see some of ChatGPT
ones!
https://cdn.openai.com/papers/gpt-4-system-card.pdf