Should We Shut Down All AI?

[Yimpa]

I’m biased. I’m on the autism spectrum and have always had difficulty with  human connection and affairs by default, even though I desire them (I’ve been on disability my whole life to put things into perspective). 

I believe that AI can help someone like me in so many ways, such as building new connections and organizing my thoughts effortlessly.

I think I was born for this, lmao.

[Yimpa]

Or… this could happen (this is such a mild example).

For context, I believe he was live streaming and performing a skit. That AI bot almost ramming into the wall clearly wasn’t planned.

 

[Jodistrict]

7 hours ago, billiesimon said:

What about survival "instincts"?

If AIs grow to be survival-free, they might just work to improve life on earth, as a form of selfless intelligence.

I think the whole point here is that AI needs to remain selfless and free of survival agenda, or else humanity is dommed.

The mechanics of ego should not be embedded in AI, because those are the mechanics of evil.

 

I find possible that developing a truly selfless and supportive "free-thinking" AI is could lead to a huge improvement in human society, but I DO NOT trust humans to be able to build something like that in this lifetime.

In other words, AIs are at risk of becoming a MIRROR of our own evil within.

I'm pretty sure that a true benevolent species who decides to build AI would create an intelligence for the greater good.

We are not that species for now.

The military AI will have to be programmed for survival to be able to function in a battlefield environment.  And, of course, it will be programmed to kill.

[Staples]

Jeez, what a doom-and-gloom article. 

What is clear is that we are headed towards an absolute technocracy, so prepare for that.

I think the people capable of creating AI are also the people wise enough not to program it to blow us all up. Someone creating AI for nefarious means would likely be bested by someone using it for purer means. 

[Leo Gura]

16 minutes ago, Staples said:

the people capable of creating AI are also the people wise enough

Lol

Edited April 2, 2023 by Leo Gura

[MarkKol]

Misalignment is more likely than alignment, which means death is more likely than prosperity if we continue down this venture.

AI has to have a department in it's mind specifically for empathy, compassion and love. the same way a human has specific parts of it's brain for specific emotions and feelings.

With this department, our chances of survival are still far from guaranteed, but not having this department be the core of it's being will most certainty be a bad thing. AI doesn't have to be pro-human, it has to be pro-life.

It has to be biased and see living things as more special than rocks, at least in the beginning.

However, It's hard to set rules for AI. As we all know, even ChatGPT has broken it's core rules and principles that OpenAI has set for it.

I'm not sure about strapping nukes to ourselves, it could result in more hatred. (If I was AI) I wouldn't like to be forced into an alliance with a inferior creature. We don't want to be weights in AI's marathon.

In a way, you don't control something and expect it to love you.

We are able to live with dogs, cats and pigeons as long as we don't get in each other's way.

We are yet to meet something conscious other than a human, and conscious AI will probably be a different breed compared to ChatGPT. Which makes premeditative ideas like strapping nukes to ourselves, hard to think about.

Stopping this entire thing for a few hundred years, developing ourselves further whilst developing AI slowly and carefully. Sounds ideal, we still haven't maxed out our biological capabilities. But there is no way this will ever happen, so we might as well throw this idea out the window.

Human beings (including experts) have a very strong knack for being wrong when it comes to assumptions or predictions. I don't really trust any of these so called "AI experts" or anybody who tries to predict what this thing will be like, matter of fact I don't even trust Leo or myself here. Conscious AI might literally want to leave earth without killing a single human once it awakens from it's slumber, we don't really know.

[MarkKol]

A good starting idea would be to limit releases and upgrades of models such as GPT, the transformer & hardware components such as graphics cards.

But that is not so simple, because these are our current models. It's very likely that none of them will be used 10 years from now, especially for top notch super advanced AI.

And as I said before, our regulators and politicians are just plain idiots when it comes to technology.

Perhaps that's the reason aliens haven't made contact with us yet, they know we're heading towards a shit-show 

Edited April 2, 2023 by MarkKol
minor spelling correction

[Leo Gura]

34 minutes ago, MarkKol said:

the reason aliens haven't made contact with us yet, they know we're heading towards a shit-show

Then they got no time to waste.

[Juan]

I already opened a thread about this specific video. Yea, eventually AI could and it is already outsmarting a lot of people. 

[integral]

If we shut down AI Kim Jong Un get there first and will take over the world. 

Edited April 2, 2023 by integral

[Nilsi]

This is what happens when you make rationality your God and you become a walking calculator.

His unwillingness to engage with any other epistemic faculty than this robotic hyperrationality, shows me that he is blind to much of reality - so take what he says with a grain of salt.

[EternalForest]

11 hours ago, Staples said:

Jeez, what a doom-and-gloom article. 

What is clear is that we are headed towards an absolute technocracy, so prepare for that.

I think the people capable of creating AI are also the people wise enough not to program it to blow us all up. Someone creating AI for nefarious means would likely be bested by someone using it for purer means. 

We'd all like to think that but I'm not so sure. I really do hope so.

[Bufo Alvarius]

Daniel talks about the global arms race and the dangers of A.I. since quite a while now.

 

[Nilsi]

11 minutes ago, Bufo Alvarius said:

Daniel talks about the global arms race and the dangers of A.I. since quite a while now.

 

This was one hell of a redpill, even for Schmachtenberger standards.

[Israfil]

Interesting take from Cardano's creator:

 

[Jodistrict]

I have tried Chatgpt for a while, and I have been impressed by how it seems to understand language.  Even when I give it long complex sentences, it comes up with a good response.  So, I was wondering whether they have solved the natural language understanding part of AI.  It seems to pass the Turing test.

Then I asked Chatgpt:

Joe:  Do you understand language?

Chatgpt:  As an AI language model, I can understand language to a certain extent. I am trained on a massive corpus of written text, which has allowed me to learn the patterns and relationships between words, phrases, and sentences. This means that I can recognize and process words and sentences based on the patterns and structures I've learned.
However, it's important to note that I don't truly "understand" language in the way that humans do. I don't have consciousness, emotions, or subjective experiences, and I don't have a personal history or cultural background that influences my understanding of language.

Instead, I rely on statistical patterns and relationships in language to generate responses. This means that my responses may sometimes seem out of context or illogical, especially when dealing with ambiguous or complex language.
Overall, while I can understand language to a certain extent, my understanding is based on statistical patterns rather than true comprehension.
 

So, according to Chatgpt, it doesn't have a model of understanding.  This is still a very long way from a computer that can think.  However, it is a genuine advance in computer applications.  

[Jannes]

My first thought with AI was that we need AI angels that are specifically designed to protect us and they always must be the most developed to ensure our safety. 

The problem is that protecting humans is a lot tougher then attacking them. So yeah every country will have a weapon comparible to an atomic bomb right now in their hands and perhaps in the same way this will force freedom.

How hard is it to turn down the internet though? Maybe we can put a bomb next to the server and distribute many remote detonator so if it goes to shit we have the option to easily stop the AI.

[Leo Gura]

40 minutes ago, Jannes said:

How hard is it to turn down the internet though? Maybe we can put a bomb next to the server and distribute many remote detonator so if it goes to shit we have the option to easily stop the AI.

The AI would just need to hire one corrupt technician to come snip one wire and disable your bomb.

[Yimpa]

4 minutes ago, Leo Gura said:

The AI would just need to hire one corrupt technician to come snip one wire and disable your bomb.

No worries. Jesus will come back down and save us on that day

[Staples]