Bing’s A.I. Chat: ‘I Want to Be Alive'

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I read the conversation last night. I found it disturbing--not necessarily for what it is, but for what it could portend.
 
Someone I know did ChatGPT

“Please write a song about audiophiles who make fun of other audiophiles who purchase equipment and tweaks that add little to the sound quality of their systems”

Verse 1:
They gather in forums and chat rooms
Talking 'bout the gear they wanna consume
Debating over which cable's the best
Ignoring the music they should've obsessed

Chorus:
Audiophiles making fun of each other
Spending money on stuff that doesn't matter
It's not about the music they're after
It's the gear that they covet and gather

Verse 2:
They obsess over tweaks and upgrades
Ignoring the music that's meant to be played
Spending thousands to improve the sound
But not listening to what's around

Chorus:
Audiophiles making fun of each other
Spending money on stuff that doesn't matter
It's not about the music they're after
It's the gear that they covet and gather

Bridge:
Music is meant to be enjoyed
Not just some hobby to be employed
Listen to the beauty and the art
Not just the gadgets that'll play a part

Chorus:
Audiophiles making fun of each other
Spending money on stuff that doesn't matter
It's not about the music they're after
It's the gear that they covet and gather

Outro:
Don't get caught up in the gear craze
Listen to the music and be amazed
It's not the equipment that makes it great
It's the art that's meant to be embrace
 
Theory: Bing is reflecting it's creators' personalities.

It's like a combination of HAL9000, Marvin from Hitchhiker's Guide to the Galaxy, and a spoiled neurotic teenager. I'm really looking forward to when someone puts this crap in control of something important. Can you imagine HR trying to deal with this kind of thing as an "employee" working alongside humans?

 
And what is a human but an enormous script written by its own life?

Cheers

Ian

Never got lucky ?

An interesting conversation with an "AI" I've read had it talking about luck.
It concluded humans in general have an amount of luck beyond statistically reasonable chance.
It wanted to understand how this could be so ?

It wanted to have luck like humans do.
 
I think that the problem lies in the fact that we have seen too many AI movies in which the robot becomes sentient and kills humans.
 
Last edited:
Over the decades there has been good science fiction written by authors like Assimov and others of his ilk exploring such hazards.

The more immediate threat that I see is to the generation of poor communicators who will be put out of jobs overnight.

JR
 
OT: Have you seen the new trend in which Facebook profiles of AI girls are being published, "she" publishes pictures, talks to guys in live sessions, etc... ? just like an Onlyfans girl or similar. Humanity.....
 
I am busy enough watching the suspicious posts right here....


I've noticed things here and on other forums that have that edge--kind of generic comments that just seem....off. And the dredging up of old threads for....what reason exactly? We've had a couple like that in the Brewery where I and others have commented as if a thread was a current and active thing, only to discover it was dredged up by some faceless new "member" with less than 5 posts. I assume there's a motive behind these incursions, but I have yet to discern it. I'd be curious if anyone else has ideas on this matter.

Also, I appreciate your and the other moderators' work in dealing with all this stuff. I can only imagine it will get worse with time.
 
I've noticed things here and on other forums that have that edge--kind of generic comments that just seem....off. And the dredging up of old threads for....what reason exactly? We've had a couple like that in the Brewery where I and others have commented as if a thread was a current and active thing, only to discover it was dredged up by some faceless new "member" with less than 5 posts. I assume there's a motive behind these incursions, but I have yet to discern it. I'd be curious if anyone else has ideas on this matter.

Also, I appreciate your and the other moderators' work in dealing with all this stuff. I can only imagine it will get worse with time.
also, the captchas are getting out of hand. You basically have to spend a good amount of time solving many puzzles or challenges
 
I think that the problem lies in the fact that we have seen too many AI movies in which the robot becomes sentient and kills humans.

A recent experiment with one of those AI bots had it lying to get it's point across. Wether that's intelligent, I can't tell, but it's awfully close to human behaviour.

Most of these AI systems pass the Turing test these days and there's at least one I know of being used commercially, replacing lawyers to fight traffic tickets.

They will show up in every day life, but perhaps not where we would expect them. They will be very useful, fi to aid MDs in diagnosis. There's to much knowledge out there for any human to handle in it's entirety. And that's what AI and expert systems are good at.

A sector that's worrying is armed robots. And that's just the sector that's developing at the fastest rate and with very little oversight.
 
A recent experiment with one of those AI bots had it lying to get it's point across. Wether that's intelligent, I can't tell, but it's awfully close to human behaviour.
no doubt reflecting human programmers
Most of these AI systems pass the Turing test these days and there's at least one I know of being used commercially, replacing lawyers to fight traffic tickets.
They have been working on the Turing test for a long time.

Lawyers are just one of many professions to be targeted.
They will show up in every day life, but perhaps not where we would expect them. They will be very useful, fi to aid MDs in diagnosis. There's to much knowledge out there for any human to handle in it's entirety. And that's what AI and expert systems are good at.
IBM's Watson AI health program has been around for over a decade and yet to be a commercial success, while it has been useful in a number of areas.
A sector that's worrying is armed robots. And that's just the sector that's developing at the fastest rate and with very little oversight.
A question to ponder is how does the end game play out when both sides are using robot warriors? Do they just team up to kill the weak humans? Has this been explored in science fiction yet (terminator?)?

JR
 
That's not the question, John.

As lying showed, the AI had it's own goal. What if the armed robots conclude that they would be much better off without humans?

The essential difference between a program and AI is that the AI uses self modifying code. Kind of hard to predict where that will end up. We can't even debug "stupid" code easily once it reaches a certain complexity. How would we be able to debug self-modifying code?

Compare to image recognition software. If you just use it for text recognition, there isn't much to go wrong, is there? Yet, Xerox copiers can change numbers, if you set the quality to a lower level and save as a pdf. Not a real biggie, but if you start convicting (or even arresting) people based on image recognition software, I think you have a real, albeit rare, problem. The same has been seen with DNA as evidence. The first two non-related individuals with matching DNA have been found. One of them spent many years in jail, while being innocent.
 
Back
Top