On 17/05/23 13:28, Gordon wrote:
https://www.stuff.co.nz/technology/digital-living/300881208/we-have-put-the-world-in-danger-with-ai-admits-chatgpt-creator
This is ggod to see that at least we are making a start to decide how to control the AI.
To me this seems like a poor thought through idea
"Eliezer Yudkowsky, a leading machine learning researcher, wrote an article in Time magazine suggesting countries should be prepared to use nuclear weapons to destroy dangerous AI projects."
Yes let us do this, splash radiation all over the globe. Probably will not wipe out AI and set the humans against each other.
We also need globally, how to, sort out what AI might do when it runs amock. Can we have a version that is socialist for China and liberty based?
Elon Musk has pointed out that he is not sure where things could end. Now that in itself is a big red (non Chineese) flag.
Some ground rules are needed as of now.
As it happens I went to a basics tutorial on ChatGPT yesterday.
Had a quick hands on, on the tutors account. I didn't create one
for myself and probably never will because it is obliviously recording everything u input. Learning on the job. Big privacy issue IMO.
I asked it to write some code in JavaScript, which it did.
Next question was 'give me 10 prime numbers.'
It didn't produce any numbers. I got JavaScript code to generate primes.
That's learning on the job, based on the previous question.
So, through usage the app will increasingly generate output based on
current memes. It will rapidly adjust to only provide narratives based
on acceptable 'truths'.
There is also a copyright issue. It will not provide source references. Probably relying on a 'Creative Commons' and 'Open Source' legal
protection. But, it also means that you can easily be given the
'official narrative' without identifying that source is controlled.
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)