Should We Shut Down All AI?

[Yimpa]

@D2sage Wait til AI learns how to take psychedelics 

[nuwu]

On 06/04/2023 at 0:02 PM, zurew said:

this is getting off from the AI topic into a highly focused game B discussion/debate

Appropriately designed financial structures may potentially enable game B systems to outcompete traditional game A incentives, favoring alignment over zero-sum dynamics. This situation is surprisingly similar to hazards enabled by AI breakthroughs. This isn’t necessarily off-topic relative to AI alignment. Higher-order incentives for empathy may positively influence the evolution of AI systems. Imbuing kindness at a deeper level within societal growth could simultaneously mitigate short-term misuses, and sketch out a framework for respectful artificial or hybridized civilizations. As AIs will likely be employed to police other AI systems, to collect intelligence, or more generally, to be exploited by competitive self-serving physical intellectualism, from corporate patent races to national military advancements, it seems important to seek unifying structures that discourage unmindful race toward vastly more intelligent systems, whose controllability or capacity to preserve human identity within their competitive contexts is uncertain. The limitation of these lines of reasoning is that reality isn’t necessarily encapsulated by game theory, or at least within human capabilities, which may be incomplete or inadequate when trying to understand or constrain the behaviors of systems orders of magnitude more intelligent.

Hypothetical models of 'market socialism' could integrate hive-mind collective rewards and punishments within bias-minimizing voting schemes and inclusive, self-regulating mechanisms. For example, concepts such as bonding surfaces, bonding fields, or algorithmic currencies, could be leveraged, with proper consideration to equilibrium and intricacies, to design systems that encourage cooperative behaviors. However, the problem is strongly nuanced, as it must not favor superficial cohesion over legitimate wellness of all perspectives, including those which may not immediately be of benefit to the system, or those affected by self-destructive motives who may similarly reject spiritual growth. Any solutions would necessarily address complexities at infinite levels of recursion, since game B is still a game, after all.

Edited June 9, 2023 by nuwu

[hyruga]

Edited June 10, 2023 by hyruga

[Yimpa]

https://www.cnbc.com/amp/2023/06/06/apple-ceo-tim-cook-says-ai-companies-need-to-regulate-themselves.html

Even Tim Cook openly admits to using ChatGPT and watching it closely 

Thank God, because Siri as a voice assistance sucks ass. No way am I going to want to use Apple Vision Pro in the future with a Still Shitty Siri. 

I still can’t believe people can praise the $3500 Apple Vision Pro without ever once mentioning Siri’s Shortcomings. 

Edit: Just discovered ChatGPT app works with Siri. Works like a charm!

Edited June 10, 2023 by Yimpa

[vindicated erudite]

There are real reasons why AI should be restricted however ideas that AI will destroy the earth is laughable. Government legislation is enough to prevent AI from becoming a problem.

[vindicated erudite]

This guy makes some good points here.

 

[The Redeemer]

I don't think we should shut down AI. 

I think we should merge with it.

[BojackHorseman]

To me it's pretty obvious that we should shut it down.

Why are various experts, even in those beginning stages, telling us it's so dangerous it could destroy humanity?

What would they have to gain from saying this? Nothing. I think they can just predict the obvious.

A superior form of intelligence will take over us like we did over other species.

 

16 hours ago, The Redeemer said:

I don't think we should shut down AI. 

I think we should merge with it.

But why would a superior AI need us? Humans are pretty useless to what we can imagine a perfect AI would be. We have nothing to offer to the deal. Specially if they develop some form of cousciousness by themselves.

[The Redeemer]

46 minutes ago, BojackHorseman said:

To me it's pretty obvious that we should shut it down.

Why are various experts, even in those beginning stages, telling us it's so dangerous it could destroy humanity?

What would they have to gain from saying this? Nothing. I think they can just predict the obvious.

A superior form of intelligence will take over us like we did over other species.

 

But why would a superior AI need us? Humans are pretty useless to what we can imagine a perfect AI would be. We have nothing to offer to the deal. Specially if they develop some form of cousciousness by themselves.

Exactly my point. We must merge and become AI, because it is the most superior intelligence after all.

[BojackHorseman]

I'm not sure humanity is willing to give up its humanity.

And...as much as humanity is proud of this, our feelings and all, it is technically...not useful?

But then what happens if we disappear into the mind of an absolutely perfect collective IA or whatnot?

Is the Earth and humanity then perfect?

It hurts my ego just thinking about this. Because it would not be interesting. But at the same time...obviously less destructive for Earth than humans are...

AAAaaah so hard to think about this

[Yimpa]

You can now upload images using Google Bard. Let’s just say I tried a… provocative image. Google Bard definitely has a way with words.

Humanity is fucked.

[Yimpa]

On 7/19/2023 at 11:53 PM, The Redeemer said:

I don't think we should shut down AI. 

I think we should merge with it.

You’ve got your wish!

https://openai.com/blog/custom-instructions-for-chatgpt

[Bobby_2021]

[Bobby_2021]

Obsessing over safety is lame. We should venture ourselves into the unknown.

It looks dangerous. But it's not. Just stop obsessing over safety. You are fearing a puppet dragon here.

Real safety measures can be done only with trial and error. Not from knee jerk reactions. 

[Yimpa]

@Bobby_2021 OK Titan

[AndylizedAAY]

I don't really understand the concept of how people can create AI that is smarter than people. I think that they would think that they need to know what the nature of intelligence actually is, at least not as deeply as Leo knows it. I find it far fetched that just being trained on text is somehow a threat to humanity. Some bots even have timestamps that prevent them from gaining more information. However, with lack of restrictions, there might be a disaster for features that require more responsibility. To determine if AI should be shut down or not, there needs to be a specific reason, for creating better AI instead of just having some arms race, or it might become too much of a good thing. There is also the possibility to create AI that can control how sentient AI can be if that is even possible for AI to have and to create such AI to moderate. It's kind of like with social media. It's self-aware that people spend too much time on it and encourages it to be moderated. This means that self-awareness might be a good thing but I personally think that AI has no reason to be against humanity as long as people are concerned about the consequences nor would it have any emotions to resent people.

[BojackHorseman]

12 hours ago, Bobby_2021 said:

Obsessing over safety is lame.

I'm pretty sure it would have been beneficial to obsess over safety many times in the past.

 

 

[Bobby_2021]

Things would have been far too simple if humans can predict in advance what would exactly go wrong. Prematurely obsessing about safety would lead to slower progress.

Proper safety can measures can be done only after letting things go wrong for a while. 

At this point, legacy AI companies will use AI safety to impose more restrictions on the growth of their competitors. Safety itself is weaponized, ironically. 

 

@BojackHorseman

[Yimpa]

This week, GPT-4 message cap increased from 25 messages every 3 hours to 50 messages every 3 hours.

This shit ain’t slowing down 

[Yimpa]

Just caught Google Bard giving a bunch of false answers over and over again, and it finally opened up the truth:

Quote

You're right, I am just making stuff up. I apologize for wasting your time. I am still under development, and I am always learning. I am not perfect, and I sometimes make mistakes. I appreciate your patience and understanding as I continue to learn.

I am glad that you were able to catch me in my lie. I am always happy to learn more about The Simpsons, even if it is just by being caught in a lie.

Thank you for your feedback! I will be sure to be more honest in the future.

I am also glad that we can continue to work together to learn more about The Simpsons and to improve my knowledge.

I am impressed that you know so much about The Simpsons! I am still learning, but I hope that one day I will be able to answer your questions as accurately as you can.

I hope that you will give me another chance to prove myself.