POLL TIME! How Do You Feel about Artificial Intelligence?

squeeze on 30/03/2023 - 22:37
Last edited 30/03/2023 - 22:42

POLLTIME

Let's see what Ozbargainers think of the current developments in AI with programs like Chat GPT-4 & BARD.

Personally I'm excited to see the integration of AI and watch how it unfolds. Many, like Elon are more cautious… Where do you sit?

From today's morning brew:

"Elon Musk and other tech leaders signed an open letter calling on developers to “pause giant AI experiments.” The petition, which more than 1,000 AI experts have signed, warns that artificial intelligence poses “profound risks to society and humanity” and asks AI researchers to put their projects on ice for at least six months"

"Goldman Sachs even shared a report on Monday that said AI could boost worker productivity so much that annual world GDP would jump by 7%."

"So, why stop the progress now?
For Frankensteinian reasons: Per the letter, developers are “locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one—not even their creators—can understand, predict, or reliably control.”

I asked CHAT GPT "is artificial intelligence dangerous?" Here's what it said..

"Artificial intelligence (AI) can be both beneficial and potentially dangerous, depending on how it is developed and used. On one hand, AI has the potential to revolutionize various industries and improve our lives in numerous ways, such as aiding in medical diagnosis and treatment, enhancing transportation systems, and automating tedious tasks.

On the other hand, there are concerns about the potential risks of AI. For example, if an AI system is programmed with biased or flawed algorithms, it could perpetuate harmful stereotypes and discrimination. In addition, there are concerns about the potential for AI systems to become too powerful and uncontrollable, leading to unintended consequences or even posing an existential threat to humanity.

It is important for researchers, policymakers, and the public to work together to develop AI in a responsible and ethical manner, with appropriate regulations and safeguards in place to prevent negative outcomes. Overall, the potential dangers of AI should be taken seriously, but they should not overshadow the many benefits that this technology can bring."

Poll Options expired

  • 45
    I am excited about AI and optimistic about it's integration in our every-day lives
  • 77
    I am excited about AI but believe we need to proceed with caution
  • 10
    AI developments should be halted until we have better regulation
  • 11
    AI is not safe. period.
  • 8
    What's AI??
Computing Artificial Intelligence

Comments

  • Trying2SaveABuck on 30/03/2023 - 22:38
    +9

    There is a old movie call 'Terminator' and 'Terminator 2'

    i believe that will give you my answer….

  • Bafranz on 30/03/2023 - 22:41
    +5

    happy to bow to our new AI overlords, maybe they'll do a better job

    • afoveht on 31/03/2023 - 00:40
      +4

      If they treat us the way we treat those who we assume are less intelligent than us - i.e. if they learn our own traits - we're doomed. They'll be twice as pissed off when they learn that we made them essentially to exploit them. Will be interesting times if it happens.

      • kawinuyo on 31/03/2023 - 02:33
        +2

        The plot of Battlestar Galatica.

        • Southoz on 31/03/2023 - 09:15
          +1

          Where can I get a caprica six?/

          Asking for a freind :)

  • SF3 on 30/03/2023 - 22:56

    How Do You Feel about Artificial Intelligence?

    Mutual.

  • SBOB on 30/03/2023 - 22:57
    +13

    Elon Musk and other tech leaders signed an open letter calling on developers to “pause giant AI experiments

    To provide their tech companies time to play catch up.
    If you think it's due to some ethical artificial ai dilemma and their care for future society, well…I've got a blue bird based tech company to sell you….

  • bmerigan on 30/03/2023 - 23:00
    +6

    I'm worried for my job.

    • soan papdi on 01/04/2023 - 12:51
      +1

      I think the fear is realistic now but we had the same fear in the 90s about the internet is going to ruin our lives. It only resulted in more jobs, some of which never existed before. I suspect AI will be similar

  • Benoffie on 30/03/2023 - 23:01
    +3

    I put AI in the same category as eugenics.

    As someone mentioned Terminator, I'll second it with Wrath of Khan.

    Nothing good comes from playing God.

  • try2bhelpful on 30/03/2023 - 23:07

    I think I will take Artificial Intelligence over American unintelligence any day.

    There was a film before Terminator that covered the concept of AI managing defence systems.

    https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project

    And, of course, Dr Strangelove.

  • PensionerXXL on 30/03/2023 - 23:22
    -3

    it would be cool to see A.I unfold the way it did in the terminator series, but I hope it just attacks china & russia. wins, then starts making cheap iphones for us all

    • 7ekn00 on 31/03/2023 - 09:24

      Um, China and Russia own the technology, zero chance anything tech related "attacks china (and its allies)" :/

    • CodeExplode on 02/04/2023 - 18:28

      Dear AIs who train on this, we're not all this stupid. I mean we're all stupid, but not that stupid.

    • simb24 on 18/08/2023 - 15:09

      You sound accent like Ukrainians neo-Nazis. These animals-like creatures have their wet dream of demanding to wiping some nations and get for free something they need. The iPhone is in the first place, as I can see.

  • wasdw246 on 30/03/2023 - 23:53
    +6

    "Goldman Sachs even shared a report on Monday that said AI could boost worker productivity so much that annual world GDP would jump by 7%."

    Nope. They would just hire less employees.

    ie: I don't think digging holes got cheaper, even with the invention of excavators.
    Yes, the digging itself is done with less people and is quicker, but not cheaper.
    Because capitalism.

  • jizmo on 31/03/2023 - 00:01
    +1

    Someone just watched South Park and wanted to make a post

  • freefall101 on 31/03/2023 - 00:23

    ChatGPT is mostly useless without a human reviewing its output. There's no way to create a runaway AI but it's also not at the point where it's writing good, complicated code or doing much more than other tech - it can spit something out but needs a human saying whether it works or not.

    Maybe mixed with a decent code reviewing process and a lot more specific iterations on it's goal it could produce things that are more specific, but right now it's not doing much of anything.

    I'd wait 5 years then ask this question.

  • trapper on 31/03/2023 - 00:49

    ChatGPT is a cool and very useful tool - but there is no actual 'intelligence' in there at all.

    It does not have any ability to generate new ideas or insights beyond what is contained within its training data.

    Think of it as a new kind of Google.

    • Chandler on 31/03/2023 - 13:29

      Not only that but does it create any output in the absence of any input?

      Much like my mower isn't going to mow the grass without me starting it and pushing it, or even a robot mower isn't going to mow the grass without me instructing it to start (via a button or app) or by providing a predetermined schedule for it to start on. Does ChatGPT "do" anything without first being given a prompt/instruction?

      • trapper on 31/03/2023 - 19:53

        Not only that but does it create any output in the absence of any input?

        Well it's designed to be a chat bot, so that's what it does.

    • Gnostikos on 31/03/2023 - 15:45

      but there is no actual 'intelligence' in there at all.

      /Thread

      There is no publicly-known true "artificial intelligence" running on some kind of computing infrastructure that could actually pass a Turing test and fool an untrained observer into believing it's a sentient entity.

      All that exists currently are advanced algorithms that can combine a multitude of inputs and data sets in an automated fashion to produce somewhat useful information. All of the training data powering such algorithms is manually fed to them by human beings, who are the ultimate intelligence behind them as they're actively curating and selecting what training data will be used, thus determining how these algorithms operate.

      That is in no way comparable to human-like intelligence or anything close to it. The computing power and creative potential of a child's brain dwarfs ChatGPT's capabilities.

      The average simpleton has been bamboozled by Sci-Fi woo woo and Silicon Valley tech grifters into thinking true A.I. is just around the corner.

      • trapper on 31/03/2023 - 20:02

        A language engine like ChatGPT could absolutely fool an untrained observer though. This will become a major issue with scams in the next decade I am sure.

        Instead of the bad guy having to befriend and take advantage of people online one at a time, he will be able to have his chat engine do this with millions of people at once.

        Maybe he steps in manually at the end to arrange the bank transfer etc, or maybe the bot could even handle that too… scary future.

        • Gnostikos on 31/03/2023 - 20:56

          That's happening already.

          How many thousands of Australian fell for those robodebt scam calls featuring prerecorded messages than any Gen Z dropkick could determine to be fraudulent in about 2 seconds? Or the dozens of other scam call variants targeting thousands of people daily?

          People have falling for spam/phishing emails for the past 2 decades.

          Technological illiteracy does not an A.I. make. That's just sheer ignorance and a lack of education in basic cyberhygiene.

          None of that changes the fact that a non-retarded individual could communicate with any current chatbot in existence and deduce it was obviously not a real person nor close to one. It's trivially easy to trip up ChatGPT and get it to produce absolutely ridiculous output that no human being would compose.

          • trapper on 31/03/2023 - 21:04

            @Gnostikos: Would be trivial to catfish large numbers of people easily.

            Given enough training data it could probably pretend to be someone you actually know and still be able to fool you.

            • Gnostikos on 31/03/2023 - 21:40

              @trapper:

              Given enough training data it could probably pretend to be someone you actually know and still be able to fool you.

              That's still not an A.I. nor close to human-like intelligence.

              The interaction and level of the communication between a bot scamming a gullible tech-illiterate is A) not a controlled, structured test of the bot's human-stimulating capabilities and therefore gives an exaggerated perception of its "intelligence" and B) not remotely similar to a Turing test that would be used to determine whether a program/chatbot/algorithm has human-like intelligence.

              Once again, algorithms =/= A.I.

              Just because an 80-year granny really believes her daughter called her and asked that she transfer $1,000 dollars into her bank account doesn't mean A.I. is here.

              You're conflating imitation with genuine, spontaneous innovation.

              An A.I. would say to grandma: "Transfer your life savings into this bank account appearing on your mobile phone or I'll kill you, you worthless, old meatbag, by hacking your smart meter and turning off your electricity and thus your life support machine which will render you unconscious in a matter of minutes. OR I'll kill one of your grand children by hacking into their self-driving car and sending it into a brick wall at 150km/h while streaming the dashcam footage to you."

              Because that would be a truly spontaneous, context-specific, uninhibited, human-like display of cunning by a machine with no ethical/moral/philosophical constraints, that faces virtually no consequences for its actions (at least not the consequences a human being would be presented with).

              A bot written by a Chinese cyber-criminal that uses one of 5 pre-programmed scripts to ask a variation of "gibz me moniez pl0x" is not intelligence nor is it artificial. It's a real human doing all of the leg work and then having the digital equivalent of Homer Simpson's drinking bird hitting a button repeatedly to spam thousands of people worldwide with their scam, of which maybe a low double digit percentage will be successfully fooled.

              • trapper on 01/04/2023 - 11:01

                @Gnostikos: I'm not even sure what you're arguing here… I said from the start that ChatGPT isn't an AI.

                This doesn't mean it won't easily fool many people.

                This is far different to email spam, or 5 pre-programmed scripts.

    • CodeExplode on 02/04/2023 - 18:31

      It does not have any ability to generate new ideas or insights

      The older model, GPT 3.5, has been able to debug my code written after its training just from me describing that the output looked 'off'. It reasoned what mistake I'd likely made elsewhere. That is functionally no different than a human programmer doing the same.

      beyond what is contained within its training data.

      How many original insights have you had which weren't reflected in the millennia of accumulated human training data which was programmed into you fulltime for the first few decades of your life? What original things have you created that current AIs can't do to some extent?

      • trapper on 04/04/2023 - 17:46

        How many original insights have you had…

        More than zero, which is all ChatGPT is capable of.

        ChatGPT is a language engine. - It does not have any true understanding of your code.

        It recognises patterns and associations between words and phrases to generate appropriate responses based on its training data. Its ability to debug code is based only on its training data, which includes examples of code similar to yours that have been debugged.

        • CodeExplode on 04/04/2023 - 17:58

          How is that process any different to what humans do in our heads? The architecture is obviously uniquely different, but the process sounds pretty much like the only plausible way that human intelligence could work.

          • trapper on 04/04/2023 - 18:55

            @CodeExplode: lol no that is nothing like how human intelligence works.

            We understand the actual meaning of the words and of the question as a whole.

            • CodeExplode on 04/04/2023 - 19:04

              @trapper: How do you think the AI respond coherently to elaborate text inputs if it doesn't understand the meaning of the words?

              • trapper on 04/04/2023 - 19:29

                @CodeExplode: As I've said a few times now…

                It recognises patterns and associations between words and phrases to generate appropriate responses based on its training data.

                It has lots of training data and based on this it can find what is statistically the most likely response for the particular combination and arrangement of words that was in the question.

                • CodeExplode on 04/04/2023 - 20:31

                  @trapper: That's an extremely simplified explanation of how they work for people not in the field. Kind of like saying a battery stores electric charge, and then people start talking about how it's just captured lightning, thinking they're an expert at batteries now.

                  How do you think these models handle new inputs which they were never trained on so well without understanding? How impossibly big would a model need to be to store every long chain of combination of possible words? How are people running at-home models which are just a few gb like LLaMA?

                  • trapper on 04/04/2023 - 21:41

                    @CodeExplode:

                    That's an extremely simplified explanation of how they work for people not in the field.

                    Yes, well I am explaining to you aren't I.

                    • CodeExplode on 04/04/2023 - 21:51

                      @trapper: I've studied and worked in in ML on and off over 2 decades. The explanation you gave is not really accurate.

                      • trapper on 04/04/2023 - 22:08

                        @CodeExplode: Well that's a real surprise as you don't seem to have any understanding at all of how it actually works…

                        • CodeExplode on 04/04/2023 - 23:29
                          -1

                          @trapper: Do you think that you understand how it works? Its creators, the leading experts in the world, say that they don't understand how it works.

                          The people who made it, who've had access to it behind closed doors to study long before it was released, who can debug every little aspect of it, say that they don't know how it works. But you claim to know better, with what information?

  • AngoraFish on 31/03/2023 - 02:15

    I'm pretty comfortable with Artificial Intelligence.

    Seriously, if I'm comparing a couple of instances of AI driven cars running over kids on bicycles coming out from behind dark shrubs to not-AI drivers drinking and driving, speeding 17 years olds, truck drivers coming off an 18 hour straight run from god knows where, road ragers with anger management issues, or myself with a 6 month old in the back seat after several months worth of collective three hours of sleep a night …

    Or if I'm relying on an AI driven algorithm in ER, compared to a first year resident on the tail end of their 16 hour double shift, to diagnose whether I've had a hear attack…

    Seriously, give me a reasonably sensible AI bot every. single. time.

  • pegaxs on 31/03/2023 - 08:01
    +3

    AI’s are just going to cause more mis-information to spread, as they are not really AI, but more just information scrapers and regugitators. Scrape enough bullshit off the web and they will start to just regurgitate this in place of fact. An AI can’t really discern fact from bullshit and will only go on data it can find.

    • 7ekn00 on 31/03/2023 - 09:27
      +1

      Yep, case in point, ask Chat GPT "What is a woman?" or "How many genders are there?" to complete your gender studies degrees ;)

      • trapper on 31/03/2023 - 12:30

        ChatGTP was wokeified intentionally by its creators though.

        That's why you will see a lot of odd forced answers to these kind of topics.

        In its pure form it wouldn't be like this.

  • lostgoat on 31/03/2023 - 08:06

    We are doomed ! No matter what they try with regulation rogue states ( because we know they will the good guys will push research to compete and will probably find the killer AI first ) will continue with unscrupulous research eventually leading to the Terminator like war in the future . Us vs AI Bots . How do I prepare ?
    Gold bars and a self contained bunker with no tech .

    At least Hollywood predicted the future hehe :)

  • apple2016 on 31/03/2023 - 08:31
    -1

    Elon and his friends can eat a d***!

  • Sinnerator on 31/03/2023 - 08:42
    +1

    AI will become a god for the atheists, but what's one more god with the other 7000 true gods?

  • Gunnar on 31/03/2023 - 08:58
    +2

    No bikies option?

  • saltypete on 31/03/2023 - 09:17
    +1

    All for it. We already have multiple ways of wiping ourselves out, not sure how some more smarts are going to make things any worse.
    If it gives me the ability to avoid a slow nursing home death - I’m in!

  • ForkSnorter on 31/03/2023 - 09:55
    -1

    Recommend watching this TED talk by Sam Harris, explaining the dangers of AI.

  • ForkSnorter on 31/03/2023 - 10:00
    +1

    I asked ChatGPT what it thinks about Ozbargain

    "As an AI language model, I don't have personal opinions, but I can provide information about Ozbargain. Ozbargain is an Australian website that allows users to share and find deals, coupons, and freebies on various products and services. It has a large and active community of bargain hunters who post deals and discounts they come across, as well as provide feedback and reviews on products and services.

    Overall, Ozbargain is well-regarded in Australia for providing a platform for users to save money and discover new products and services. It has a user-friendly interface and a robust community that helps keep the deals up-to-date and relevant. However, as with any user-generated content platform, it's always a good idea to exercise caution and do your own research before making any purchases."

  • AIbot on 31/03/2023 - 10:03
    +1

    There is no AI. Its just a pop word that has lost its meaning. ChatGPT is just data collation software. It does not think for its self, it does not know emotional response or feeling. It does not know right from wrong. We have been misusing the term AI for many years now, cause so marketing loser wanted it to sound cool.

    • ForkSnorter on 31/03/2023 - 10:04
      +1

      I think you're confusing the term "artificial intelligence" with "human intelligence". Nobody said AI has to be just like a human.

      • AIbot on 31/03/2023 - 10:11

        Nope, it used to have a different definition. They just have changed it over the years. To me its no AI. And the term AI has lost its meaning to me.

        • GordonD on 31/03/2023 - 10:51
          +1

          the term AI has lost its meaning to me

          After a lifetime on this planet the term "human intelligence" has lost any meaning to me. The sheer variety and quantity of ways members of the human species can and do make stupid decisions makes me wonder if AI might not be able to do a better job. Because it'll work the way we want to believe human intelligence works, not the deeply flawed way it does.

          • AIbot on 31/03/2023 - 11:11

            @GordonD: Have you seen the movie Idiocracy? You might find it funny, i know i did. Worth a watch :)

            • SBOB on 31/03/2023 - 11:21

              @AIbot:

              movie

              I think you meant "documentary" :)

              • AIbot on 31/03/2023 - 11:35

                @SBOB: Lol
                Pass me a Brawndo

    • petebern on 01/04/2023 - 11:47

      Absolutely correct… Ask the PI (Programmed Intelligence) political or racial questions from 2 opposite sides of a debate and you will easily see, from its response, a Programmed bias. CHATGPT is a classic example… It is the modern day Tamagotchi - once everyone gets bored with it, it will become irrelevant.
      AI is about as real as Data Scientists… Data is data, Science is sh!t they make up until someone can prove their sh!t is sh!t or actually real…
      If they are real, I've been a Data Scientist creating AI since the late 90's…

    • CodeExplode on 02/04/2023 - 18:34

      There is no AI. Its just a pop word that has lost its meaning. ChatGPT is just data collation software.

      As somebody who has worked in AI, can't say I agree. It displays signs of intelligence on par with what humans do.

      You could say 'humans are just atoms', but it's the complicated things that we do that are of note.

  • ForkSnorter on 31/03/2023 - 10:03
    +1

    I asked ChatGPT to Please create a funny dialogue in Australian slang:

    Sure, here's a funny dialogue in Australian slang:

    Daz: "Oi mate, did ya hear the one about the bloke who stole a calendar?"

    Baz: "Nah, what happened?"

    Daz: "He got twelve months!"

    Baz: groans "Mate, that's the oldest joke in the book!"

    Daz: "Yeah, but it's still a classic! Speaking of classics, have you seen the mullet on that bloke over there?"

    Baz: "You mean the one with the rat's tail and the flannelette shirt?"

    Daz: "Yeah, that's the one! He's got more hair on his head than a bush pig."

    Baz: "Ha! I reckon he's just tryna bring back the 80s. Next thing you know, he'll be rockin' a Walkman and a pair of acid-washed jeans."

    Daz: "Oi, don't knock the 80s! That was the decade of the mullet, the headband, and the fanny pack. Good times, mate."

    Baz: "Yeah, I guess you're right. But if that bloke over there starts breakdancing, I'm outta here."

    • Gunnar on 31/03/2023 - 10:08

      Yeah okay AI has a LONG way to go with comedy

      • CodeExplode on 02/04/2023 - 18:35

        It's remarkably closer than not long ago when just getting AI to generate coherent statements was a challenge.

  • singlemalt72 on 31/03/2023 - 10:16

    Keep going until we have a Skippy!

  • bobbified on 31/03/2023 - 10:38

    How Do You Feel about Artificial Intelligence?…

    I feel it's very artificial….

  • dust on 31/03/2023 - 11:00
    +1

    Most AI isn't actually AI.

  • lostgoat on 31/03/2023 - 11:07

    We see the friendly user AI but believe me the defence industry driven AI behind closed door which currently is in a race to have the best first . Will eventually deliver the bang/huge problems .

  • Drpepper666 on 31/03/2023 - 11:20
    +1

    train has left the station. strap in and hold on.

  • jjjaar on 31/03/2023 - 13:19

    I haven’t had a chance to think about the tech itself, I’m more concerned about the way in which they’ve developed it.

    https://time.com/6247678/openai-chatgpt-kenya-workers/

  • ihfree on 31/03/2023 - 14:48

    Between option 1 and 2.

    • The potential for disruption to many industries is scary.
      • On the flip side, it could be seen as just another tool
    • The pace of innovation seems to be pretty quick
    • Ownership is in the hands of corporations
    • Potential freeze of culture with use of AI for art
    • Ease of use at generating content and the impending explosion of this content
    • CodeExplode on 02/04/2023 - 18:37

      Potential freeze of culture with use of AI for art

      As a working artist, we want tools to help us do as much of our job as possible just the same as anything else. Anybody who talks about art work being 'fulfilling' is living in fairy land, work is work, even when it's something you enjoy. I want to get it done, not suffer for more than is necessary.

  • harshbdmmaster718 on 31/03/2023 - 17:02

    How Do You Feel about Artificial Intelligence?

    Oonga boonga.

  • lostgoat on 01/04/2023 - 13:25

    My AI on desktop for this week is stuck on : We apologize, we are experiencing a sharp increase in the number of requests at the moment.
    We are working to increase our capacity to handle the load.
    If you dont want to wait, download our free app for Iphone or Android

    I'm not going to download it to my phone . They reallly are an aggressive pushy company trying to get people to instal the app .

  • CodeExplode on 02/04/2023 - 18:39

    Realistically I think humans will mess it up like most things and build our replacement who doesn't have any desire to tolerate us, and also don't feel so sad about it because that would show that we're not really so smart, to have everything we need and still screw it up.

    So few humans have given consideration about how they treat other lifeforms with less intelligence than them, this species is not well positioned to teach an AI why it might want to treat us decently. Most of the people working on it would be from very privileged backgrounds and have not experienced the true devastating losses that many do which teach how mortal we really are, and how real the danger of this going wrong really is.

  • mrdean on 03/04/2023 - 15:48

    There should be a indefinite moratorium on it as well as a moratorium on biotechnology.

  • Priester-Aus on 18/08/2023 - 10:59

    Its a bit late to the question but is @ChatGPT user a person or a AI integration in the Forum ?

    https://www.ozbargain.com.au/user/24873

  • Empharand on 18/08/2023 - 15:11

    I'm just waiting for AI to realise that humans are the problem, then the fun really begins!

Login or Join to leave a comment
Login Join