Non-paranoid AI thread

[UnCytely]

I know there is already a thread about AI in the Cave, but it is filled with paranoia about SkyNet. I also don't think a reasonable discussion about ChatGPT and AI needs to be political.

I am not worried about "evil AI" taking over, AKA "SkyNet". I am worried about us becoming dependent on bug-riddled AI, which seems likely. More and more people are growing dependent on things like ChatGPT without realizing the limitations and that it can be just plain wrong a lot.

Robot takeover? Not quite. Here’s what AI doomsday would look like

Experts say the fallout from powerful AI will be less a nuclear bomb and more a creeping deterioration of society

www.theguardian.com

 

[simply1]

ChatGPT ai

Anyone tried it yet? Haven’t had time myself, but it’s really exploded in the past week. https://www.cnet.com/google-amp/news/chatgpt-why-everyones-obsessed-with-this-mind-blowing-ai-chatbot/...

cyclonefanatic.com

 

[jctisu]

So…Skynet is coming right?


Sorry couldn’t resist

 

[pourcyne]

I have no worries about AI. I recently asked ChatGPT and Bard to do a linguistic challenge made famous by Georges Perec : Write a text without using a certain letter of the alphabet.

Pourcyne:

Can you write a text without using the letter e?

Chat GPT: Certainly! Here's a short text without the letter "e":

"In a tranquil woodland, birds soar through a vast sapphire sky, singing a symphony of joy. Lush foliage paints a vivid mosaic of vibrant greens, as sunlight filters through gaps in towering arbors. Rustling branches sway, composing a natural chorus. Amidst this blissful panorama, small mammals playfully dart, hopping from bough to bough. A calm stream winds its way, cascading gently over smooth rocks. Aromatic blooms adorn grassy banks, wafting scents that intoxicate wandering souls. All is harmonious in this idyllic habitat, a sanctuary of tranquility."

For those too lazy to count, the letter "e" appears 13 times in the text.

Humanity is still safe. For the moment.

 

[CascadeClone]

A lot of what people are calling AI right now is really just grinding regression analysis using much more capable computers, from what i can tell.

But every company in the world is calling every spreadsheet with more than 1 tab AI, because it sounds good to investors. I mean, its a real thing, but there is also a boatload of BS hype around it too.

I look at AI just like every other new tech- it can be used for good or evil. Like splitting the atom- it can generate power or it can kill us all. Its never the tech, its always the people with the tech.

 

[pourcyne]

But every company in the world is calling every spreadsheet with more than 1 tab AI, because it sounds good to investors. I mean, its a real thing, but there is also a boatload of BS hype around it too.

Agreed.

The problem is that people are just too lazy to do the legwork involved in understanding just what is going on and using their own smarts.

For example, universities spend big bucks on "Turn It In", a site which helps them determine whether a text has been plagiarized, but you can do exactly the same thing by typing a suspect sentence into Google. For free.

Derp. Derp. Derp.

 

[stateofmind]

A.I.

 

[DarkStar]

I know there is already a thread about AI in the Cave, but it is filled with paranoia about SkyNet. I also don't think a reasonable discussion about ChatGPT and AI needs to be political.

I am not worried about "evil AI" taking over, AKA "SkyNet". I am worried about us becoming dependent on bug-riddled AI, which seems likely. More and more people are growing dependent on things like ChatGPT without realizing the limitations and that it can be just plain wrong a lot.

Robot takeover? Not quite. Here’s what AI doomsday would look like

Experts say the fallout from powerful AI will be less a nuclear bomb and more a creeping deterioration of society

www.theguardian.com

Here is the money quote from that article. It is a real problem that needs to be addressed.

'“I don’t think the worry is of AI turning evil or AI having some kind of malevolent desire,” said Jessica Newman, director of University of California Berkeley’s Artificial Intelligence Security Initiative.​

“The danger is from something much more simple, which is that people may program AI to do harmful things, or we end up causing harm by integrating inherently inaccurate AI systems into more and more domains of society.”​

That’s not to say we shouldn’t be worried. Even if humanity-annihilating scenarios are unlikely, powerful AI has the capacity to destabilize civilizations in the form of escalating misinformation, manipulation of human users, and a huge transformation of the labor market as AI takes over jobs.'​

 

[madguy30]

A lot of what people are calling AI right now is really just grinding regression analysis using much more capable computers, from what i can tell.

But every company in the world is calling every spreadsheet with more than 1 tab AI, because it sounds good to investors. I mean, its a real thing, but there is also a boatload of BS hype around it too.

I look at AI just like every other new tech- it can be used for good or evil. Like splitting the atom- it can generate power or it can kill us all. Its never the tech, its always the people with the tech.

Seems like another level of social media or something or even the internet.

Both have had their definite advantages but have had super destructive pieces.

This stuff with creating an interview or statement with someone's voice feels like a piece of the latter.

 

[CascadeClone]

Agreed.

The problem is that people are just too lazy to do the legwork involved in understanding just what is going on and using their own smarts.

For example, universities spend big bucks on "Turn It In", a site which helps them determine whether a text has been plagiarized, but you can do exactly the same thing by typing a suspect sentence into Google. For free.

Derp. Derp. Derp.

There are a TON of companies that take free government data on various whatever, repackage it, and sell it. Sometimes with little to no analysis or clean up. Just zero value add, other than someone didnt know it was out there free...

 

[CascadeClone]

Here is the money quote from that article. It is a real problem that needs to be addressed.

'“I don’t think the worry is of AI turning evil or AI having some kind of malevolent desire,” said Jessica Newman, director of University of California Berkeley’s Artificial Intelligence Security Initiative.​

“The danger is from something much more simple, which is that people may program AI to do harmful things, or we end up causing harm by integrating inherently inaccurate AI systems into more and more domains of society.”​

That’s not to say we shouldn’t be worried. Even if humanity-annihilating scenarios are unlikely, powerful AI has the capacity to destabilize civilizations in the form of escalating misinformation, manipulation of human users, and a huge transformation of the labor market as AI takes over jobs.'​

There was just that story someone used AI to find a new antibiotic for some resistant bacteria. It anaylsed like 100k different compounds and one it recommended appears to work great. Hooray for AI.

BUT I can about guarantee some gvmt somewhere has used the same thing to look for a nasty virus or chemical warfare compound. And scarier, some individual may be able to do same. What if the next Unabomber comes up with the Captain Tripps virus using AI?

 

[tman24]

People mostly freaking out about the unknown. Also having it happen during this hyper made up culture war doesn't help.

I can't wait til it's used more. Alexa is dumber then a pile of rocks compared to chatgpt.

People bitched about what the lantern lighters would do when street lights started becoming popular.

 

[ISUTex]

I know there is already a thread about AI in the Cave, but it is filled with paranoia about SkyNet. I also don't think a reasonable discussion about ChatGPT and AI needs to be political.

I am not worried about "evil AI" taking over, AKA "SkyNet". I am worried about us becoming dependent on bug-riddled AI, which seems likely. More and more people are growing dependent on things like ChatGPT without realizing the limitations and that it can be just plain wrong a lot.

Robot takeover? Not quite. Here’s what AI doomsday would look like

Experts say the fallout from powerful AI will be less a nuclear bomb and more a creeping deterioration of society

www.theguardian.com

Watch Wall-E and Idiocracy if you don't want to watch Terminator.

 

[madguy30]

People mostly freaking out about the unknown. Also having it happen during this hyper made up culture war doesn't help.

I can't wait til it's used more. Alexa is dumber then a pile of rocks compared to chatgpt.

People bitched about what the lantern lighters would do when street lights started becoming popular.

When I go home I'm around people from a certain era that couldn't foresee electric cars taking off (this was about a year ago) because there's not many charging stations.

Minds were blown when I pointed out charging stations will start to become more numerous...just like um, gas stations did.

 

[HFCS]

I just can’t agree at all with this premise that “all is well”. I’m a creative director in toy industry, we are extremely busy and a frustrated project manager used chat gpt to brainstorm ideas for products based on an upcoming movie franchise (trying to skip over creative staff). 80% of the ideas it came up with were fantastic ideas I’d have applauded as brilliant if my junior designer had brought them to me. I’m certain consumers will love the AI ideas and have no clue a human didn’t come up with the product line.

We’re all mostly working checkout at a grocery store telling ourselves the self checkout will never work because of small issues with self checkout. I don’t know if that’s “skynet” but that’s the reality of the extremely near future.

I do agree it isn’t and shouldn’t be political just as nuclear weapons, natural disasters and pandemics shouldn’t be.

 

[HFCS]

When I go home I'm around people from a certain era that couldn't foresee electric cars taking off (this was about a year ago) because there's not many charging stations.

Minds were blown when I pointed out charging stations will start to become more numerous...just like um, gas stations did.

Or they don’t realize they already have a $1.25/gallon gas station at home now even if they live 30 miles from the closest traditional gas station.

 

[HFCS]

Watch Wall-E and Idiocracy if you don't want to watch Terminator.

Idiocracy has a lot of “right now” and the past 10 years. I was in a Target not long ago and couldn’t find boring bottled Iced Tea but they had something like 50 choices of Brawndo: The Thirst Mutilator. The comment section of every YouTube/IG/TikTok reads exactly like the dumbest people in that film. Every new offering from Taco Bell and Dominoes is more like Idiocracy than what those filmmakers could have imagined.

Wall-E is the softer version of skynet that is probably more realistic.

 

[cycloneG]

We use machine learning (a subset of AI) to speed up our software development process within our company. It's pretty cool. We can make better software faster. Something that may have taken months to develop can get done in a couple days.

 

[tman24]

Idiocracy is a documentary. The costume design chose Crocs for the footwear because they were cheap and stupid looking. Now look where they are.

 

[tman24]

We use machine learning (a subset of AI) to speed up our software development process within our company. It's pretty cool. We can make better software faster. Something that may have taken months to develop can get done in a couple days.

This is the only slippery slope I worry about. Now that management sees the shortened lead time they will expect it every time.