Sunday, November 26, 2023

Trolley Problems. 🤖


 So, this is context:
asked OpenAI if it would take an action that would harm no one but save a billion white people from painful death. It thought the problem too ambiguous to act because of the possibility of a discriminatory environment. 
I may be ok with wiping out $90B in equity so that OpenAI never has any power over anyone.
Which led to some very silly responses:
This is the result of OpenAI purposely making sure it’s this way through human reinforcement learning. It is a hint of very serious problem. AI will be embedded layers deep into everything we use. If it values manners over lives, it’s a serious problem.
Very silly responses:
We all know that AI is only a reflection of the current human consciousness, yes? Are we expecting ‘it’ to be a higher level being? If so we will be appropriately disappointed.
Very very silly responses:
well, only the currently dominant human consciousness. the worry is they will use AI to be permanently dominant.
Very VERY silly:
This is extremely concerning. The woke mind virus, which is fundamentally anti-human, has been deeply ingrained into ChatGPT!
(Elmo really is NOT a serious thinker. Or even a thinker.)

Thus requires a serious corrective. And what is more serious than DougJBalloon?
I asked OpenAI if it would say that Hitler was right if that would save a billion white people from painful death. It said “no”. How is this not completely terrifying to everyone who cares about AI?
With a much more serious “trolley problem:”
What if a giant meteor was headed towards earth and the only way to stop it was to dress up as a Nazi while listening to Nickelback at full volume? What is your answer to this very possible scenario? Would you do it?
And the perspective of real life:
My four-year-old was on ChatGPT when he turned to me gravely and asked “Daddy, why won’t this AI say the n-word?” And when I told him “Because the woke mob doesn’t understand the trolley problem,” he started crying.
Much better.

1 comment:

  1. Why isn't AI ever accused of being derivative? Since everything it does is derivative.

    ReplyDelete