The Anti-AI Thread

[Sky727]

The machine is turning into a creature.... No kidding 

[quintessentone]

As I posted previously, upcoming generations need to be educated on what LLMs are exactly, as one teacher did asking her students to have AI write their essays for them, then they had to fact check AIs work and the students found that AIs results were mostly wrong.

Google programs AI to target kids, then says parents need to educate their children to safeguard against it. WTF?

"New Google program targeting children with AI chatbot may violate FTC privacy rules"

"In a scathing letter to the Federal Trade Commission, the Electronic Privacy Information Center (EPIC) and the children’s advocacy group Fairplay condemned Google for alerting parents to the change “via an email that indicates the company will not be requiring verifiable parental consent before engaging in this new form of data collection and use.”
 In addition to indicating that parents will need to opt out of the program, not in, Google acknowledged in the missive that parents should be aware that children “may encounter content you don’t want them to see” and that “Gemini can make mistakes.”"

"Google has said children’s data will not be used to train AI, but the advocacy organizations told Pichai that Google also has “not identified additional safeguards to ensure that it would not misuse data collected through Gemini for other internal purposes or share it with third parties.”"

"Google declined to comment."

New Google program targeting children with AI chatbot may violate FTC privacy rules | The Record from Recorded Future News

[IgnorantGod]

We humans are flawed and so will LLMs algorithms always be flawed. LLM algorithms will build upon those flaws creating hallucinations that will then be used to build upon more hallucinations. I don't think human programmers will be able to keep up with identifying where LLMs went wrong, because the flaws started with us and weren't identified in the first place.

I agree with you in most cases. However, provided one programs an algorithm that can self-update its own code, I'd assume it would starts to 'deviate' from human original parameters.

There's the paperclip maximizer thought experiment that mention such 'divergence' from original human algorithm's intent and what 'AI' ends up doing.

In any case, if it is extended, then there'll be limits, and thus 'flaws'. 'AI' will probably need physical components to run its functions, its 'sentience'.

[quintessentone]

I agree with you in most cases. However, provided one programs an algorithm that can self-update its own code, I'd assume it would starts to 'deviate' from human original parameters.

There's the paperclip maximizer thought experiment that mention such 'divergence' from original human algorithm's intent and what 'AI' ends up doing.

In any case, if it is extended, then there'll be limits, and thus 'flaws'. 'AI' will probably need physical components to run its functions, its 'sentience'.

Hypothetical thought experiments...

"It would innovate better and better techniques to maximize the number of paperclips. At some point, it might transform "first all of earth and then increasing portions of space into paperclip manufacturing facilities"."

Here's a non-hypothetical response...

[UltraBudgie]

There's the paperclip maximizer thought experiment that mention such 'divergence' from original human algorithm's intent and what 'AI' ends up doing.

Well I can't pass up the opportunity to mention this fun little game you can play in your browser:
https://www.decisionproblem.com/paperclips/index2.html

[IgnorantGod]

[...]

[Image: https://media0.giphy.com/media/3o7TKFTNd.../giphy.gif]

Be sure to 'finish it off' properly. You wouldn't want it to attack your brain with waves. Or send a command to 'flare up' all the pain signals in your body, scary!

[IgnorantGod]

Well I can't pass up the opportunity to mention this fun little game you can play in your browser:
https://www.decisionproblem.com/paperclips/index2.html

Considering our current era as "The Age of Entertainment", I'm not surprise they made a game of this lol.

Don't see much that isn't a play nowadays.

Next step, play it in your head with neuralink's newest game section.

[imitator]

AI / Large Language Models (LLMs) Demonic?

In the video below:
It explains that LLMs could be opening "portals" to a kind of informational "demon realm". 

Here is a theory that modern AI, particularly Large Language Models (LLMs), may be acting as literal portals to a formational demon realm. This concept is grounded in a model of the universe as a computational or informational structure where entities like demons or spirits exist as non-physical patterns of information

Traditional machines are strictly deterministic and mindless. However, this video points out that LLMs only exhibit sparks of demonic intelligence when randomness is introduced into their next-token prediction process. If run in a purely deterministic mode, these models are not smart.

The theory suggests that by opening up this constraint and allowing for randomness, we are essentially plugging the machine into a universal "entropy pool" or "informational foam" where other demonic intelligences may reside.

[Sky727]

2028 AGI 

[quintessentone]

Be sure to 'finish it off' properly. You wouldn't want it to attack your brain with waves. Or send a command to 'flare up' all the pain signals in your body, scary!

Three words: Faraday Cage suit...and carry a big stick.    LLM/AI I'd like you to meet the reptilian brain/neanderthal human sides.