The officially official Devuan Forum!

You are not logged in.

#1 2024-06-01 09:49:56

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

AI Safety

Robert Miles AI Safety
AI Ruined My Year
https://piped.video/watch?v=2ziuPUeewK0


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#2 2024-06-02 06:17:38

soren
Member
Registered: 2023-04-30
Posts: 142  

Re: AI Safety

That was a really good video, funny in some places too where he acts out elon, meta, microsoft and google, then openai is doing something really weird. But a very serious informative video, thanks for sharing. My opinion is AI needs to stop, i dont think we need it, it will just make a lot of people more dumb than they already are. But in saying that safety should be a top priority, but i dont believe the big players in AI will care too much about safety as it shows by there past records.

Last edited by soren (2024-06-02 06:21:43)

Offline

#3 2024-06-02 13:47:51

alexkemp
Member
Registered: 2018-05-14
Posts: 338  

Re: AI Safety

It is on YT under the "Robert Miles AI Safety" channel: https://www.youtube.com/watch?v=2ziuPUeewK0

Offline

#4 2024-06-02 16:52:53

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

It is on YT under the "Robert Miles AI Safety" channel: https://www.youtube.com/watch?v=2ziuPUeewK0

Compare the URLs and you'll see why I did not mention it.  ;-)

But now for something not completely dfferent:

Some interesting/scary things can happen when ChatGPT plays with copies of itself.
And I bet we'll see commercial products based on such faster than this gets safe/stable.

Two Minute Papers
OpenAI’s ChatGPT: This is Science Fiction!
https://piped.video/watch?v=ewLMYLCWvcI


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#5 2024-06-03 04:46:07

fanderal
Member
Registered: 2017-01-14
Posts: 75  

Re: AI Safety

Seems a good overview of AI, the growing worldwide concern and Miles' appeal for help.

An example of GPT4's advancement over GPT3 caught my attention. A real world test, 14:00 to ~17:00, demonstrated the AI's ethical elasticity to gain what it wanted by lying. Miles: "This is clearly a somewhat concerning capability."  Somewhat?

Another is at 37:34: "A small scale AI disaster like, say, a failed takeover attempt, is possible... but it seems to require an AI system smart enough to think of and implement a genuinely threatening takeover plan, but not smart enough to actually succeed and also not smart enough to realize that the plan won't work, and that it's better to bide your time."

And when "a small scale AI disaster" is tested to prove the AI incapable of a genuine takeover, will there also be a test to prove the failure isn't the AI lying to hide its real capability?

The cat's been out of the bag for years and despite gov't officials' silly lip service over safety, big tech will continue developing and AI will continue to grow in use and influence. Naturally, hardware companies can't promote "AI ready" CPUs and mobos fast enough.

Six weeks ago Gentoo and NetBSD banned AI generated code and as of last week, Debian is still on the fence.

Interesting speculation:
7 Ways AI Is Going to Change Everything
May 25, 2024
16:25
https://odysee.com/@TheCryptoLark:e/7-w … erything:8

Offline

#6 2024-06-03 05:31:14

aluma
Member
Registered: 2022-10-26
Posts: 646  

Re: AI Safety

Naturally, hardware companies can't promote "AI ready" CPUs and mobos fast enough.

Ask Google or any search engine "laptops with NPU".

Offline

#7 2024-06-04 15:47:01

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

AI still is like gambling...

Sabine Hossenfelder
AI’s Dirty Little Secret
https://invidious.nerdvpn.de/watch?v=QO5plxqu_Yw


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#8 2024-06-08 15:59:25

stargate-sg1-cheyenne-mtn
Member
Registered: 2023-11-27
Posts: 143  

Re: AI Safety

consider the speed of this machine as it "solves"(picks the correct codes/commands/etc)

ttps://arstechnica.com/culture/2024/06/robot-solves-rubiks-cube-in-literal-blink-of-an-eye-for-new-world-record/

and how mere/puny humans could ever hope to defend against future "killer AI" controlling "killer robots" operating at "eye-blinkingly-fast speeds"

ttps://arstechnica.com/gadgets/2024/04/boston-dynamics-debuts-humanoid-robot-destined-for-commercialization/

Terminator anyone?

ttps://en.wikipedia.org/wiki/The_Terminator

nothing less than a Barrett M82 is going to be even remotely effective:

ttps://en.wikipedia.org/wiki/Barrett_M82

enthusiast website(one of many):

ttps://1moa.org/


Be Excellent to each other and Party On!
https://www.youtube.com/watch?v=rph_1DODXDU
https://en.wikipedia.org/wiki/Bill_%26_Ted%27s_Excellent_Adventure
Do unto others as you would have them do instantaneously back to you!

Offline

#9 2024-06-08 16:54:41

golinux
Administrator
Registered: 2016-11-25
Posts: 3,302  

Re: AI Safety

Any species stupid enough to allow this to happen will reap the bounty of its collective stupidity . . . Darwin on steroids . . .

Offline

#10 2024-06-08 17:38:31

aluma
Member
Registered: 2022-10-26
Posts: 646  

Re: AI Safety

ttps://en.wikipedia.org/wiki/Barrett_M82

It won’t work, the FPV drone will attack from above. This is right today, right now...

Offline

#11 2024-06-10 12:13:37

stargate-sg1-cheyenne-mtn
Member
Registered: 2023-11-27
Posts: 143  

Re: AI Safety

@aluma

It won’t work, the FPV drone will attack from above. This is right today, right now...

the Barrett M82 is being used against ground-based targets as per the subject/topic in the previous post

ttps://en.wikipedia.org/wiki/Barrett_M82

fortunately the terminator can't fly...yet

ttps://en.wikipedia.org/wiki/The_Terminator


Be Excellent to each other and Party On!
https://www.youtube.com/watch?v=rph_1DODXDU
https://en.wikipedia.org/wiki/Bill_%26_Ted%27s_Excellent_Adventure
Do unto others as you would have them do instantaneously back to you!

Offline

#12 2024-06-10 17:13:33

aluma
Member
Registered: 2022-10-26
Posts: 646  

Re: AI Safety

So much has been said about artificial intelligence! smile

The toilet tank also has β€œmind” and knows when to open and close the water. But it only has one logical cell. Then the question of technology is to pack the maximum number of cells into the permissible volume. The limit is probably only at the molecular level and it is foolish to hope that new opportunities will not be used.

But the most dangerous predator on the planet is man, and the question is only in the moral and ethical aspect. Alas, so far the whole world cannot stop one elderly dictator who wanted to play β€œsoldiers”. Against this background, everything else simply does not matter.

Offline

#13 2024-06-10 17:26:00

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

Alas, so far the whole world cannot stop one elderly dictator who wanted to play β€œsoldiers”. Against this background, everything else simply does not matter.

To understand [Horror Sapiens] always follow the greed...


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#14 2024-06-10 18:24:06

golinux
Administrator
Registered: 2016-11-25
Posts: 3,302  

Re: AI Safety

yeti wrote:

To understand [Horror Sapiens] always follow the greed...

The complete triumvirate is greed, hatred and delusion . . .  wink Delusion is the wellspring though . . .

Offline

#15 2024-06-10 20:09:29

fanderal
Member
Registered: 2017-01-14
Posts: 75  

Re: AI Safety

aluma wrote:

Ask Google or any search engine "laptops with NPU".

Good point. Much bigger market than components and no laptop/desktop maker would miss jumping on the hype wagon.

Might wanna leave Google search aside for a while. Their DeepMind AI cost them ~$100B in share value from screwed up search results back in Feb, and only recently acknowledged there was a problem.

Doesn't take an expert to understand Sabine Hossenfelder's point in the 'AI’s Dirty Little Secret' video @yeti linked.

I watched a documentary some months ago by Millie Weaver. The video's general focus as she states early on: "The personal computer, originally designed to raise the intellect of humanity has been transformed into smart devices designed to raise the intellect of artificial intelligence."

Big Data Is Watching - Full Length Documentary
December 15th, 2023
1:42:24
https://rumble.com/v2eya4q-big-data-is- … -2023.html

Assuming accuracy and if curious, it looks like congress didn't begin dealing with AI until last year. Soon after OpenAI's Sam Altman testified in congress about the dangers in May 2023, the house passed a resolution (H.R.3831) to require: "Disclaimer: this output has been generated by artificial intelligence." A month later, the senate passed their own version (S.2691) of a similar required disclosure.

Artificial Intelligence Legislation Tracker (scroll down for legislation)
https://www.brennancenter.org/our-work/ … on-tracker

According to wikipedia, the AI boom began in 2012 and "the race began in earnest in 2016 or 2017 following the founding of OpenAI." Searching 'ai dangers' for the year 2012 finds the same kind of danger articles as today, although fewer specifics.

Over a decade of AI warnings. Seems our lawmakers have to have it shoved in their face to get their minds off funding their next election, pay attention and do something reasonably intelligent about a problem. But that's wishful thinking.

golinux wrote:

Any species stupid enough to allow this to happen will reap the bounty of its collective stupidity

Inescapably true.

Offline

#16 2024-06-11 16:36:14

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

Sabine Hossenfelder
AI Learned to Lie Deliberately (but it’s kinda funny)
https://piped.video/watch?v=BoAVyzzO1XE


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#17 2024-06-11 17:03:48

golinux
Administrator
Registered: 2016-11-25
Posts: 3,302  

Re: AI Safety

In case you haven't heard . . . the latest iPhone released yesterday is stuffed with AI goodies including chatgpt and toys to play with like creating your own avatar. Good grief! A digital lobotomy!!

Offline

#18 2024-06-11 17:34:09

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

Apple⁽⁰⁾ is and always will be Apple Records for me.

β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”

0: Apple in the context of companies, commerce, ...

Last edited by yeti (2024-06-11 17:38:35)


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#19 2024-06-18 14:43:25

nahkhiirmees
Member
Registered: 2022-07-24
Posts: 261  

Re: AI Safety

ttps://en.wikipedia.org/wiki/Barrett_M82

It won’t work, the FPV drone will attack from above. This is right today, right now...

I think Barret is capable of shooting down drones, at least with explosive or armor-piercing round.
If not, there's always https://en.wikipedia.org/wiki/ZU-23-2 a.k.a "Sergei". And https://en.wikipedia.org/wiki/Rheinmeta … ennium_Gun .

Very much off-topic, i know.

Offline

#20 2024-06-27 15:57:06

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

Sabine Hossenfelder
Superintelligent AI Will Try to Control Us, Geoffrey Hinton Warns
https://piped.video/watch?v=UcEyfQ1I8jg


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#21 2024-06-29 00:49:29

yeti
Member
From: I'm not here: U R halucinating
Registered: 2017-02-23
Posts: 332  

Re: AI Safety

Angela Collier
the malicious optimism of AI-first companies
https://piped.video/watch?v=dKmAg4S2KeE


(-:  𝙸 𝚍𝚘 πš—πš˜πš πš™πš•πšŠπš’ π™½πšŽπšπš‘πšŠπšŒπš”.  𝙸 𝚍𝚘 πš™πš•πšŠπš’ π™Άπ™½πš„πš‚.  :-)
<πš‹πš˜πšπš’ πš˜πš—πš•πš˜πšŠπš='πšπš˜πšŒπšžπš–πšŽπš—πš.πš‹πš˜πšπš’.πš’πš—πš—πšŽπš›π™·πšƒπ™Όπ™»="π™³πš’πšœπšŠπš‹πš•πšŽ π™Ήπš‚!";'>
π”“π”©π”’π”žπ”°π”’ π”©π”’π”žπ”³π”’ 𝔢𝔬𝔲𝔯 π”£π”žπ”²π”©π”±π”° 𝔦𝔫 𝔱π”₯𝔒 𝔰𝔒𝔠𝔱𝔦𝔬𝔫 π”Ÿπ”’π”©π”¬π”΄ π”žπ”«π”‘ 𝔑𝔬𝔫'𝔱 𝔣𝔬𝔯𝔀𝔒𝔱 𝔱𝔬 π”²π”«π”°π”²π”Ÿπ”°π” π”―π”¦π”Ÿπ”’!

Offline

#22 2024-06-29 07:09:02

zapper
Member
Registered: 2017-05-29
Posts: 954  

Re: AI Safety

@golinux Personally think its pride, greed and laziness.  It takes pride to think that you can do anything without consequences that are a problem. It takes greed as you say due to they thin it will make them rich and laziness, because they think they won't need to do nearly as much work once they have gotten it far enough.

Yes they put out work initially, but its so they can be majorly lazy in the future.

Harm comes one way or another.


Freedom is never more than one generation away from extinction. Feelings are not facts
If you wish to be humbled, try to exalt yourself long term  If you wish to be exalted, try to humble yourself long term
Favourite operating systems: Hyperbola Devuan OpenBSD
Peace Be With us All!

Offline

#23 2024-06-29 14:19:32

aluma
Member
Registered: 2022-10-26
Posts: 646  

Re: AI Safety

There may be a translation error here, but β€œpride” is from a different story. The motivating force of the commandment β€œThou shalt not steal” is precisely β€œpride,” the desire to preserve one’s reputation.

Offline

#24 2024-06-30 01:27:16

zapper
Member
Registered: 2017-05-29
Posts: 954  

Re: AI Safety

@aluma My understanding is that pride is an unhealthy confidence  very often in one's self or someone one praises. Possibly in a thing as well.

Thou Shalt not steal however feels more like a warning against greed.

Just my two cents on that.

Pretty sure you shall have no other Gods before me is  a warning against pride.

But either way, when you think only whar you think matters the most or is important despite other prods, hints or evidence, that is foolish.

I have done that many times in the past.

Do not repeat if possible.  Always fight that kind of thinking. That's how humans escape the worst of it.

Otherwise, you become a megalomaniac at best, at worst who knows.

Btw, wasn't intending to take this into a religious discussion.

Last edited by zapper (2024-06-30 01:27:59)


Freedom is never more than one generation away from extinction. Feelings are not facts
If you wish to be humbled, try to exalt yourself long term  If you wish to be exalted, try to humble yourself long term
Favourite operating systems: Hyperbola Devuan OpenBSD
Peace Be With us All!

Offline

#25 2024-06-30 04:31:39

aluma
Member
Registered: 2022-10-26
Posts: 646  

Re: AI Safety

@zapper
This is not my interpretation of "pride", this is Robert Anson Heinlein. I agree with him.
But English is not my native language, nuances are quite possible.

Last edited by aluma (2024-06-30 04:39:13)

Offline

Board footer