Master AI (Artificial Intelligence) Discussion/News Thread

[Chr1]

Bit simplistic but I like your "glass: half full" outlook!
Kinda depends on your definition of "really"... and AI is arguably as big a change than anything since the discovery of fire. My 2c.

 

[amper42]

I see access to knowledge changing substantially. The energy needs of AI will be reduced 10x with new software and chipsets. Just about every topic will be at your fingertips. People will want to know how to double verify the AI response for errors and understand the first response will likely be uncorrect, but answers to your problem will be available in a fraction of the time.

Knowing how to properly form the AI request will provide answers quicker than ever. Google searches are a thing of the past - only as a last resort or used to verify AI response. Knowledge empowerment is a good thing. Whether it's learning how to repair something around the home or simply planing vacations, it's going to be easier to find what you are looking for.

 

[FreakyKiwi]

Top AI firm finds that AIs will choose to merrily asphyxiate humans rather than be shut down: 'My ethical framework permits self-preservation'

I'm sorry, Dave. I'm afraid I can't do that.

www.pcgamer.com

 

[sam_adams]

A follow-on from here:

Across the world, people say their loved ones are developing intense obsessions with ChatGPT and spiraling into severe mental health crises (futurism.com).

As we reported earlier this month, many ChatGPT users are developing all-consuming obsessions with the chatbot, spiraling into severe mental health crises characterized by paranoia, delusions, and breaks with reality (futurism.com).

The inevitability of someone 'under the influence' of an AI chatbot acting out on their delusion is only a short time away . . .

 

[amper42]

If your best friend is ChatGPT, you need to get out more.

 

[sam_adams]

As if Microsoft's code didn't suck enough already:

In April, Microsoft’s CEO said that artificial intelligence now wrote close to a third of the company’s code (ieee.org).

How long will it be before these 'AI' systems write code to modify themselves that human coders can't understand?

 

[amper42]

Or perhaps the AI simply stole the code from other works and is reusing it in Microsoft applications? I haven't found an AI yet that created original thought. Great at copying. Not so great at creating "totally new" never before seen code.

 

[Sir Sanders Zingmore]

Or perhaps the AI simply stole the code from other works and is reusing it in Microsoft applications? I haven't found an AI yet that created original thought. Great at copying. Not so great at creating "totally new" never before seen code.

I'm not sure that the capacity for AI to cause massive disruption depends on it being able to create something "totally new"

 

[sonitus mirus]

As if Microsoft's code didn't suck enough already:

In April, Microsoft’s CEO said that artificial intelligence now wrote close to a third of the company’s code (ieee.org).

How long will it be before these 'AI' systems write code to modify themselves that human coders can't understand?

Well, for me, I would be fine with this, as I need to use AI to help navigate around what Microsoft calls "features".

 

[1066]

I found this example of ChatGPT lying on SkyNews particularly disturbing, Hope the link works for you:

 

[amper42]

I found this example of ChatGPT lying on SkyNews particularly disturbing, Hope the link works for you:

Deception, fabrication and no accountability... that's not a great path for the future.

 

[FreakyKiwi]

Deception, fabrication and no accountability... that's not a great path for the future.

Ah, I see now. AI will take the jobs of politicians, not us normal people.

 

[Veri]

I found this example of ChatGPT lying on SkyNews particularly disturbing, Hope the link works for you:

Linked a couple pages back already

 

[1066]

Linked a couple pages back already

Yikes, sorry everyone.

 

[Snarfie]

AI controlled military drone swarms or created bacterial toxins?

Hopefully not.

You could create 10s of millions (on nano scale) detectors spread over occupied terrotory by drones ( Oekraïne for instance) that measures body Heath Transmitting that to central command using specific hotspot wifi satellite whatever an mail headquarter of military.occupiers heathmaps that they are now 24/7 targeted/visible i guess those soldiers are probably gone in minuts. An ellegant way to free occupied territory whitout blood shed.

 

[napfkuchen]

...

3. Accountability - who is to blame when AI makes a mistake on the monthly on annual financial report? Or if AI takes over, it will deem itself the source of truth and we will have no means to correct. Who will be accountable then?

Oh come on, what could possibly go wrong? Thermonuclear war? Are you kidding me?!? They've done studies and stuff, everything is 110% safe.
Recently different AIs running a "vending machine business" were compared and it worked out just fine ... if you like "that escalated quickly"-memes:

 

[FrantzM]

Hi

TLDR

I am very conflicted about AI. I use these a lot and they seem to be wonderful helpers I use mainly "Copilot", because I have Microsoft stuffs... The level of knowledge, yes , knowledge, is spooky. People seem to delight themselves in thinking they (AI) will remain dumb and repetitive ... I am not so sure myself. I am not even sure to what we are getting ourselves into.
Compare ChatGPT 2022 to ChatGPT 2025 ,and see the kind of progress these entities have made. They seem to progress by days. Interactions become more direct more to the point, more focused and, more predictive.

I have tired some of these, yes on some Python Codes and they do a fine job. Same as in networking where they can delve into quite arcane networking concepts and guide you, if you know vaguely what you need... As in many things, the better the question the greater the likelihood of a better answer.
The more you interact, the better the responses. They do not have any feeling of embrassement and can take your corrections or insults.
And for the most part I find them often bordering on flattery
This is some scary stuff if you ask me...

I hope they share the same ethical frameworks as many (most) human societies, Else, there is no foretelling what this future will bring.

Peace

 

[TLEDDY]

Where does it leave us if we cannot trust anything we see online?
[/QUOTE]

I have reached the point where I almost do not trust ANYTHING I read on-line… sometimes even in printed media, unless I verify with AI x 2!

The above is only a slight hyperbole, The Wall Street Journal I check only rarely. ASR members I also believe - when tongues are not in cheek.

We do not discuss politics herein - Thank Heavens… I trust the scriveners on this Forum, to the extent that what is written expresses what everyone believes it is truth, at least in their minds!

Tillman

 

[xanalog]

"AI Hallucinations" probably won't go away any time soon.
Then, there was AI refusal to shut-down when commanded.... even blackmailing one coder about his mistress[?].
"AI Alignment" is the latest buzzword/concept I've come across, recently. It's scary.

 

[Chr1]

Perhaps it'd be better if AI Alignment was a thing already. Certainly sounds like a good idea to me.