[steveworkings3]

Yesterday's headlines included :

AI can pass the medical licensing exam
AI can pass the bar
AI can write an article in moments that takes a Washington Post reporter hours or days

So, doesn't this indicate that AI has gotten to or pretty close to the tipping point? And it has no morals, no conscience. I don't need to make any suggestions to the audience here what problems this could cause, and how soon they may be atop us.

I laid awake several hours last night in dread and can't find any reasons this a.m. to be more optimistic.

 

[escherm]

Didn't happen on August 27th 1997, not really worried right now to be honest.

Edit: I wouldn't be opposed to AI taking over journalism. We might be able to get some actual researched, intelligent and unbiased articles instead of useless, agenda pushing propaganda or articles about what the the fotm actor/celebrity/althete is doing.

 

[usmc124]

AI is artificial. It just speeds things up and has many practical uses like eleminating dumb errors, always available, expanding new ideas, decision making and risk reductions, repetitive jobs etc. Yes some jobs will be eliminated but many more will be added. Plus you can call a robot anything you want (no pronouns, gender bending nonsense etc.) without retribution. No reason to lose any sleep over it; its part of evolution. People lost sleep over all kinds of technology advancements and after some growing pains it becomes mainstream.

 

[Clifffalling]

ChatGPT is a problem solver for sure. I would say it is likely that what we saw in Beta release was only a segment of it's capability.
Code is code, but if the AI is allowed to rewrite it's own base code, it could be trouble.
'spose we'll see.
If you're really worried, get a GSD and train it to detect cybernetic beings.

 

[cleversloth]

Its hard to think about AI and how it works. When a person can pass the BAR or a medical exam people usually think that person is smart they studied hard. Which is usually true and usually means they are capable in other areas as well. When as AI passes the same exam that just means its able the understand enough of the questions well enough and has the data (google searches) to come up with the right answer. That's it, there is no implication that it's good at anything else or even capable of doing anything more.

AI is also really good at writing code. For now that means puking out websites and news articles but someday that may mean revising its own source code to be more efficient and add new capabilities. That's when the human race is in trouble. When an AI can really learn that's going to be a problem for us. I don't mean learn like learn a new word. I mean learn in a fundamental problem solving way. An example of when to be really concerned is when AI starts solving math problem's that graduate researches are working on today. The reason that's so significant is because if its a research problem that implies we don't know the right answer and the AI can not simply read a webpage and spit it back at you, it had to "think" to the solution based on other info all by itself to come up with something new.

read new things and say them back based on prompts (pattern recognition) = not scary at all
read new things, evaluate the information, and come up with a new unique, not yet known solution (find a new pattern) = very very scary

AI is a major concern to me but I think we're still around 10 years away. I could be wrong through, all it takes is one major breakthrough to change everything.

 

[charliemeyer007]

Wake me up when robots are designing and building robots. I think I will gladly take AI over a CONgress with no intelligence.

 

[usmc124]

Wake me up when robots are designing and building robots. I think I will gladly take AI over a CONgress with no intelligence.

He said - No Intelligence

 

[SeabeeOnce]

In its current form, ChatGPT is over-rated. It makes a lot of mistakes, still "learning". It's just code running off a rules engine, and those rules are constantly refined. I do automation for a living, and we've never replaced any humans with it. We take on the issues where mistakes are made by humans and automate it. We also free the team's time to do other tasks. Due to the financial risk, we'd never 100% automate anything. But we also automate enough that during boom times, we delay new hires, so during bust we have the money to limit layoffs.

I'm not concerned with AI, I'm more concerned with our usage of it. It pushes those users farther away from basic problem solving skills. Similar to how Snowflake pushed so many DBAs away from actually understanding Indexing and other basic skills. They simply drag and drop now, letting the backend do the thinking for them. Same for Bootstrap on the web side, very few know basic JavaScript anymore. Even network guys just use a tool to drag and drop, they don't have a firm grasp of the TCP/IP stack, at least the new ones I interact with. It's that loss of knowledge, turning "tech" workers into business users, that concerns me. Kinda like the movie Idiocracy, nobody understood the basics of anything.

 

[Cigars]

Yesterday's headlines included :

AI can pass the medical licensing exam
AI can pass the bar
AI can write an article in moments that takes a Washington Post reporter hours or days

So, doesn't this indicate that AI has gotten to or pretty close to the tipping point? And it has no morals, no conscience. I don't need to make any suggestions to the audience here what problems this could cause, and how soon they may be atop us.

I laid awake several hours last night in dread and can't find any reasons this a.m. to be more optimistic.

I was reading somewhere that AI wad a woke leaning in some of its output. So yeah, there's reason to worry. At some point, AI will reach a point where it will be considered dependable and accurate for diagnosis regarding life, death, legal matters. At that point it will be considered sentient and have protected status and will start replacing doctors, lawyers, judges, teachers, etc. Disregarding it will be considered a crime. Considering how far AI has come in the last five years alone, just think about what the next 30 years will look like.

 

[Jack Swilling]

Elon Musk considers AI to be The Existential Threat
Think about how it would be so easy for AI to create diabolical viruses, bacteria, mold, prions, and fungi. Any one or all five could easily and radically depopulate the planet.
Once robots and machines are controlled by AI there is an incredible number of really bad things that could happen.
AI access to the internet could smoke an incredible number of things.
Yada yada yada

 

[scaatylobo]

I read "I Robot" a long LONG time ago.

Then I saw the movie and I have no doubt that it could happen.

Oh and then we had " The Terminator " too.

 

[Wydaho]

Absolutely... AI is the first major step towards transhumanism. ... Yuval Harari on Transhumanism and The Singularity | YouTube - they are taking Kurtzweil's' science fiction into reality.

 

[steveworkings3]

I think Nick Bostrom is worth a listen. I've professionally written software for 35 years so I understand the call/response of what's going on right now. But it seems to me this will move faster and faster by its very nature. And the last few weeks have shown that. When the genie's out of the bottle, it'll happen quickly and few people will even know or understand. "Too late" is coming.

 

[Littlebits]

The day they use AI to control WEAPONS is the day we will have issues... as stated... AI does NOT have a conscience... it will act on what it's creators think is right or wrong... would you trust the US goverment with an AI controlling the nukes? North Korea? India? Russia? China?

 

[AK_Red]

Wake me up when robots are designing and building robots. I think I will gladly take AI over a CONgress with no intelligence.

Agreed!

I do not find an increase AI nearly as threatening as the decline in RI (real intelligence).

 

[UserNameTaken]

I'm trying to write an apocolypse novel. I've been testing a new AI program my wife found as an editor on the first chapter. It rewrites text according to the style you request, i.e., "Rewrite this text in the style of Ayn Rand, or Hemmingway." It's been interesting, mostly helpful in improving my grammer, and providing better word selection. It's nothing an educated human editor couldn't do though.

What I do is put both my text and several AI versions side by side, then cherry pick the words or text that improves what I wrote without appreciably changing my style. Some of what it comes up with is really bad, but some is quite good. Over all I think it's improved my text. We'll see what you guys think if I ever get it done.

 

[ActionJackson]

Yesterday's headlines included :

AI can pass the medical licensing exam
AI can pass the bar
AI can write an article in moments that takes a Washington Post reporter hours or days

So, doesn't this indicate that AI has gotten to or pretty close to the tipping point? And it has no morals, no conscience. I don't need to make any suggestions to the audience here what problems this could cause, and how soon they may be atop us.

I laid awake several hours last night in dread and can't find any reasons this a.m. to be more optimistic.

AI could be a big threat. It WILL be a huge threat in the hands of the wrong people. There are folks who think they know what's best for mankind and when they have full control over the various spy agencies (FBI, CIA, NSA, Interpol, etc.) then you can rest assured that AI will be one of their biggest and most well-used tools.

 

[ActionJackson]

I'm trying to write an apocolypse novel. I've been testing a new AI program my wife found as an editor on the first chapter. It rewrites text according to the style you request, i.e., "Rewrite this text in the style of Ayn Rand, or Hemmingway." It's been interesting, mostly helpful in improving my grammer, and providing better word selection. It's nothing an educated human editor couldn't do though.

What I do is put both my text and several AI versions side by side, then cherry pick the words or text that improves what I wrote without appreciably changing my style. Some of what it comes up with is really bad, but some is quite good. Over all I think it's improved my text. We'll see what you guys think if I ever get it done.

You ought to have your novel written after Shakespeare's style. "Alas, thy personal social score shalt be in the hands of the King's men, and thou shalt be surveilled roundabout."

 

[usmc124]

You ought to have your novel written after Shakespeare's style.

King James Version; I like it.

 

[SeabeeOnce]

I think Nick Bostrom is worth a listen. I've professionally written software for 35 years so I understand the call/response of what's going on right now. But it seems to me this will move faster and faster by its very nature. And the last few weeks have shown that. When the genie's out of the bottle, it'll happen quickly and few people will even know or understand. "Too late" is coming.

My Director mentioned it last week, I'm sure we'll do some kind of pilot project once we get out of this spending freeze. If the recession doesn't get me, this could eventually. Maybe I'll volunteer to be on the pilot project, just in case. I need 8 more years to pay off this house.