The absurdly alarmist reporting about AI

Discussion of cultural, religious, political or irrational subjects of any type, such as UFOs, wacko cults, mad dictators, horrible cult bands, ridiculous publications, whatever
Post Reply
User avatar
Boink Boink
Sucks Fan
Posts: 137
Joined: Thu Apr 27, 2023 8:50 pm
Been thanked: 85 times

The absurdly alarmist reporting about AI

Post by Boink Boink » Sun May 07, 2023 9:41 am

This is getting ridiculous. It's a given these days that ordinary reporters at supposedly highbrow newspapers are utter fuckwits, but now even pieces advertised as a detailed treatment of a topic are serving up pure tripe, blatantly alarmist nonsense.

https://www.theguardian.com/technology/ ... w-about-ai
The uses sound quite benign. Why are experts linking AI to the end of humanity or society as we know it?!

Perhaps a future version of ChatGPT, for instance, decides that the best way it can help people answer questions is by slowly manipulating people into putting it in charge. Or an authoritarian government hands too much autonomy to a battlefield robotics system, which decides the best way to achieve its task of winning a war is to first hold a coup in its own country.
They cite no expert for this of course.

That shite is a "reliable source" for Wikipedia, their best, even though to anyone who knows the subject as I do, it is complete garbage.

There isn't a single fucking piece of the technology behind current or future ChatGPT that does or ever will understand a concept like "in charge" or "manipulate", much less the sort of chain of consequences to achieve a given aim given in the second scenario.

Human beings are born with brains a quadrillion times more powerful than these neural networks, and it still takes years to train a human to the point it understands such thing and can, for example, consciously manipulate the person in charge. Crying of course being a less evolved instinctive behaviour.

Scientists have almost no understanding of how the human brain achieves these higher functions, chiefly because they don't really understand how consciousness arises. They do understand how brains process language, hence their success in crudely emulating it with software.

The time to shit your pants therefore, is when a scientist says they understand consciousness in humans, and then a company announces they have emulated true consciousness in a machine. You really can't do this by accident, it's not penicillin.

But for very fucking obvious reasons, this won't be done overnight, the claim will probably be massively overblown, and this won't be done in a way the artificial "mind" can do dangerous things while it is still being built and tested, which by definition includes the training stage. It certainly isn't going to be switched on at 5 o'clock and have decided by 5.01 that it has to destroy humanity.

There will be plenty of time to act, and Indeed plenty of time to ponder whether we need to act.

Why?

BECAUSE WE ARE CONSCIOUS BEINGS.

ChatGPT has no fucking clue of the significance of a question like "Should ChatGPT be destroyed before it kills humanity?" It sees no difference between that and "Should I have cheese for lunch?". There will never be a future version of ChatGPT that does. Just like there will never be a future version of Windows that can make cheese.

Proof for fuckwit journalists, if it are needed, that it is not and never will be conscious.

This alarmist nonsense is not just because it is the stuff of nonsense, but because the understandably eye catching AI IS GOING TO KILL US ALL aspects of this reporting is actually distracting people from the parts which identify the actual very real threats posed by where the technology is at right now.....
Can I trust what a chatbot tells me?
ChatGPT, Bing and Bard have produced factual errors, or hallucinations as they are known in the industry jargon. For instance, ChatGPT falsely accused an American law professor of sexual harassment and cited a non-existent Washington Post report.

the plausible nature of the responses can trick users into thinking a response is 100% correct.

There are also concerns that the technology behind chatbots could be used to produce disinformation at a significant scale. Last week, Microsoft’s chief economist warned that AI could “do a lot of damage in the hands of spammers with elections and so on”.
This should not be a mere "concern". It should be causing a three alarm fire.

We are already seeing it on a small scale with Wikipedia, a large platform vulnerable to disinformation and deliberate hoaxes where casual readers not knowing who writes what is considered normal, the preserve of the coders, but whose content is seen by some as plausibly correct, despite massive amounts of media coverage that laid out exactly what it is and therefore why you can't trust it.

Switch "Wikipedia" to "newspaper I trust", and the world is truly fucked.

Ignore these concerns, and we will have destroyed ourselves decades (centuries!?!) before any human creates a machine that is conscious as we know the term and then apparently decides the best way to survive is kill all humans.

User avatar
Bbb23sucks
Sucker
Posts: 1337
Joined: Fri Jan 06, 2023 9:08 am
Location: The Astral Plane
Has thanked: 1255 times
Been thanked: 263 times

Re: The absurdly alarmist reporting about AI

Post by Bbb23sucks » Mon May 08, 2023 12:07 pm

Boink Boink wrote:
Sun May 07, 2023 9:41 am
This is getting ridiculous. It's a given these days that ordinary reporters at supposedly highbrow newspapers are utter fuckwits, but now even pieces advertised as a detailed treatment of a topic are serving up pure tripe, blatantly alarmist nonsense.

https://www.theguardian.com/technology/ ... w-about-ai
The uses sound quite benign. Why are experts linking AI to the end of humanity or society as we know it?!

Perhaps a future version of ChatGPT, for instance, decides that the best way it can help people answer questions is by slowly manipulating people into putting it in charge. Or an authoritarian government hands too much autonomy to a battlefield robotics system, which decides the best way to achieve its task of winning a war is to first hold a coup in its own country.
They cite no expert for this of course.

That shite is a "reliable source" for Wikipedia, their best, even though to anyone who knows the subject as I do, it is complete garbage.

There isn't a single fucking piece of the technology behind current or future ChatGPT that does or ever will understand a concept like "in charge" or "manipulate", much less the sort of chain of consequences to achieve a given aim given in the second scenario.

Human beings are born with brains a quadrillion times more powerful than these neural networks, and it still takes years to train a human to the point it understands such thing and can, for example, consciously manipulate the person in charge. Crying of course being a less evolved instinctive behaviour.

Scientists have almost no understanding of how the human brain achieves these higher functions, chiefly because they don't really understand how consciousness arises. They do understand how brains process language, hence their success in crudely emulating it with software.

The time to shit your pants therefore, is when a scientist says they understand consciousness in humans, and then a company announces they have emulated true consciousness in a machine. You really can't do this by accident, it's not penicillin.

But for very fucking obvious reasons, this won't be done overnight, the claim will probably be massively overblown, and this won't be done in a way the artificial "mind" can do dangerous things while it is still being built and tested, which by definition includes the training stage. It certainly isn't going to be switched on at 5 o'clock and have decided by 5.01 that it has to destroy humanity.

There will be plenty of time to act, and Indeed plenty of time to ponder whether we need to act.

Why?

BECAUSE WE ARE CONSCIOUS BEINGS.

ChatGPT has no fucking clue of the significance of a question like "Should ChatGPT be destroyed before it kills humanity?" It sees no difference between that and "Should I have cheese for lunch?". There will never be a future version of ChatGPT that does. Just like there will never be a future version of Windows that can make cheese.

Proof for fuckwit journalists, if it are needed, that it is not and never will be conscious.

This alarmist nonsense is not just because it is the stuff of nonsense, but because the understandably eye catching AI IS GOING TO KILL US ALL aspects of this reporting is actually distracting people from the parts which identify the actual very real threats posed by where the technology is at right now.....
Can I trust what a chatbot tells me?
ChatGPT, Bing and Bard have produced factual errors, or hallucinations as they are known in the industry jargon. For instance, ChatGPT falsely accused an American law professor of sexual harassment and cited a non-existent Washington Post report.

the plausible nature of the responses can trick users into thinking a response is 100% correct.

There are also concerns that the technology behind chatbots could be used to produce disinformation at a significant scale. Last week, Microsoft’s chief economist warned that AI could “do a lot of damage in the hands of spammers with elections and so on”.
This should not be a mere "concern". It should be causing a three alarm fire.

We are already seeing it on a small scale with Wikipedia, a large platform vulnerable to disinformation and deliberate hoaxes where casual readers not knowing who writes what is considered normal, the preserve of the coders, but whose content is seen by some as plausibly correct, despite massive amounts of media coverage that laid out exactly what it is and therefore why you can't trust it.

Switch "Wikipedia" to "newspaper I trust", and the world is truly fucked.

Ignore these concerns, and we will have destroyed ourselves decades (centuries!?!) before any human creates a machine that is conscious as we know the term and then apparently decides the best way to survive is kill all humans.
I can't wait for the AI winter that will start in 6 months when the whole industry collapses.
"Globally banned" since September 5, 2023 for exposing harassment.

Email: wikipediasucks@disroot.org

Petition to ban Bbb23Wikipedia AlternativeDonate to help French strikers

Post Reply