Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI

Very Few People Are Using 'Much Hyped' AI Products Like ChatGPT, Survey Finds (bbc.com) 275

A survey of 12,000 people in six countries -- Argentina, Denmark, France, Japan, the UK, and the USA -- found that very few people are regularly using AI products like ChatGPT. Unsurprisingly, the group bucking the trend are young people ages 18 to 24. The BBC reports: Dr Richard Fletcher, the report's lead author, told the BBC there was a "mismatch" between the "hype" around AI and the "public interest" in it. The study examined views on generative AI tools -- the new generation of products that can respond to simple text prompts with human-sounding answers as well as images, audio and video. "Large parts of the public are not particularly interested in generative AI, and 30% of people in the UK say they have not heard of any of the most prominent products, including ChatGPT," Dr Fletcher said.

This research attempted to gauge what the public thinks, finding:
- The majority expect generative AI to have a large impact on society in the next five years, particularly for news, media and science
- Most said they think generative AI will make their own lives better
- When asked whether generative AI will make society as a whole better or worse, people were generally more pessimistic
In more detail, the study found: - While there is widespread awareness of generative AI overall, a sizable minority of the public -- between 20% and 30% of the online population in the six countries surveyed -- have not heard of any of the most popular AI tools.
- In terms of use, ChatGPT is by far the most widely used generative AI tool in the six countries surveyed, two or three times more widespread than the next most widely used products, Google Gemini and Microsoft Copilot.
- Younger people are much more likely to use generative AI products on a regular basis. Averaging across all six countries, 56% of 18-24s say they have used ChatGPT at least once, compared to 16% of those aged 55 and over.
- Roughly equal proportions across six countries say that they have used generative AI for getting information (24%) as creating various kinds of media, including text but also audio, code, images, and video (28%).
- Just 5% across the six countries covered say that they have used generative AI to get the latest news.

This discussion has been archived. No new comments can be posted.

Very Few People Are Using 'Much Hyped' AI Products Like ChatGPT, Survey Finds

Comments Filter:
  • Target Customers (Score:5, Insightful)

    by jythie ( 914043 ) on Thursday May 30, 2024 @05:02AM (#64509629)
    Kinda makes sense. The target audience for Ai isn't users, but investors, executives, and people hoping to resume pad.
    • Kinda makes sense. The target audience for Ai isn't users, but investors, executives, and people hoping to resume pad.

      You know what young people ages 18 to 24 are also often viewed as in the race between AI development and human employment?

      The Expendable Class.

      Go figure they're perhaps not too keen on interacting with their replacement.

      When the ChatGPT CEO plug-in is fully developed, we’ll probably find exactly zero CEOs using it. Especially when they know there’s no way in hell it could be more expensive than they are.

      • Re: (Score:3, Insightful)

        by Rei ( 128717 )

        You read it backwards. 18-24 year olds are the ones *bucking the trend*, aka using AI a lot. It's Gen X and boomers who are most mad at it.

        "But the study, from the Reuters Institute and Oxford University, says young people are bucking the trend, with 18 to 24-year-olds the most eager adopters of the tech."

        Averaging across all six countries, 56% of 18–24s say they have used ChatGPT at least once, compared to 16% of those aged 55 and over.

        To be more specific: 56% at least once or twice, 39% at least m

        • by christoban ( 3028573 ) on Thursday May 30, 2024 @06:06AM (#64509737)

          I use it several times daily. I must be like 9 years old!

          On the plus side, when Windows 12 comes out with the new AI "features," I have decided to finally just jump ship to Linux, if humanly possible. Taking constant screenshots of my PC and sending my every action to Microsoft is more than even I can take.

          In fact, I expect a significant uptick in Linux desktop use over the next few years, based solely on fear of feeding the Singularity (and Microsoft).

          • Tell us all how you use AI in your daily life. What are your USE Cases ?

            I think most people are still trying to understand The Value Proposition ... the "what's in that new tech for me?" & the "why should I use this?" questions.

          • Go for it. The water is warm. My last Windows installation disappeared about two years ago. Linux has its own quirks, of course, but it is at least as good as Windows. And doesn't put ads in your start menu.

            Mint with Cinammon is an easy start. Personally, I like Xubuntu.

            • I've stuck with Windows for decades despite loving the idea of Linux, for a few reasons:

              • Linux, with X, has had massive issues with DPI and NVIDIA, and general general bugginess. I bought an AMD PC with AMD GPU just for Linux, and it still doesn't work well on any of themany distros I've tried the last few years. I'm desperately hoping this changes before I'm forced off Windows 10, the most ergonomic OS ever. I keep saying "Wayland will help," but adoption has been so slow.
              • My eyes are poor, and incredibl
        • I stand corrected. Thanks for the clarification.

        • I am one of the outlier Boomers, I am not angry at it, I simply try each new successive generation of a few of the tools and find it less useful for me, so I don't use it, The last time I tried ChatGPT, it still was about the same (to me) as Eliza. While Eliza was a huge breakthrough in the 60's, that was 60 years ago.

          Anyway, I would like to see what they call AI become useful, and at the same time, I wish the generation between about 18 and 30 could have good jobs and such, so that they could actually af

        • Of course they are they are the ones that need to write essays for their courses.

      • by Targon ( 17348 )
        The real people who will get displaced quickly are those who do not have SKILLS. Unskilled workers are always the first to be displaced when any new breakthrough technology hits the market.
    • by NoWayNoShapeNoForm ( 7060585 ) on Thursday May 30, 2024 @07:20AM (#64509931)

      Kinda makes sense. The target audience for Ai isn't users, but investors, executives, and people hoping to resume pad.

      Like blockchain ? We don't hear much about that anymore. I guess it's no longer the hipp business tech, eh?

    • by Luckyo ( 1726890 )

      The target is actually engineers, creatives and generally people that design and build civilization.

      Most people don't build the civilization. They merely exist within it, and within its boundaries and rules. They have no need for civilization building tools. They interact not with them, but with downstream products of those tools.

      For example, almost no one interacts with the foundation of modern civilization - large scale generation of electricity. There are some that dabble in artisanal small scale generat

  • where do i use it? (Score:4, Informative)

    by fluffernutter ( 1411889 ) on Thursday May 30, 2024 @05:06AM (#64509631)
    I still don't even know where i can go to use it other than some site that wants to harvest my phone number.
    • https://chatgpt.com/ [chatgpt.com]
      Create an account with a random email account
      Phone number isn't required

      • Well, no, not giving my email address either. It did take a question without it but when I asked it to write a qt6 application it gave me "NetworkError when attempting to fetch resource".
        • Well, no, not giving my email address either.

          Wait, "your" email address? Just generate a random throwaway email account for this shit. Do we need to hand hold you on the internet?

      • Lol and now it's giving me qmake instructions even though they switched to cmake a couple years ago.
    • phind.com [phind.com]

      In the background it seems to use ChatGPT but also some other things. Most importantly, it provides references to web sites the answer is based on, so you can dig into it further on your own volition.

      It has a free variant without login that eventually has some limitations (higher query limit, more advanced models, longer follow up discussion, ...). But I find it very well suited for the occasional question now and then.

  • Plot twist (Score:4, Interesting)

    by bjoast ( 1310293 ) on Thursday May 30, 2024 @05:09AM (#64509635)
    People are using without knowing.
  • by Lavandera ( 7308312 ) on Thursday May 30, 2024 @05:10AM (#64509641)

    I use it sometimes as google search substitute...

    The problem is that ChatGPT is so often obviously wrong that I cannot really treat it seriously...

    • It may be totally wrong often, but it's still extremely useful for that 'initial draft.' And when it's wrong, you usually notice it easily. IMO, it's a lot like using a search engine and the web results it leads to: if you know its limits, it becomes a nearly indispensable tool.

    • Comment removed based on user account deletion
      • by Rei ( 128717 )

        Yes, factually, LLMs give you the consensus of the input,

        This is simply not true.

        First off, learning happens overwhelmingly in creating the foundation, not the finetunes. The foundations are unsupervised continuation models. The model learns to predict what it's going to see next in the text. This is entirely context-dependent. In the context of whether vaccines cause autism, if the context is "some rando's post on Twitter", it might be just as likely to think that they cause autism as that they don't,

        • The problem with this explanation is, there is next to no context in the foundation. Context is extremely expensive to create, and that is why it's only used in the fine tuning. There simply is no way to cost effectively provide context on the huge mass of text used for the foundation.

          And the fine tune is a lot less sophisticated than your explanation. It's simply a pile of text where metadata has been applied by the cheapest labor force available. It's most definitely not created by scientists and professi

          • by Rei ( 128717 )

            Text is its own context. Models are trained on somewhere between several thousand and several million tokens at once. That's a bloody lot of context. And trainers can add in any additional context they want into the mix.+

            And the fine tune is a lot less sophisticated than your explanation. It's simply a pile of text where metadata has been applied by the cheapest labor force available.

            You're literally talking to someone who creates her own models. *facepalm*

            I'll repeat: the finetune is "a curated dataset

  • by Spacejock ( 727523 ) on Thursday May 30, 2024 @05:11AM (#64509643)
    Not just unimpressed, the results were laughable. In three cases where I asked for information on books published in the seventies it either returned incorrect info, suggested books where the authors hadn't even been born in the 1970s, or recommended titles that had absolutely nothing to do with my query.

    At this stage it's worse than useless - instead of admitting it doesn't know something, it'll spout a likely-sounding word salad. When you call it out on the results, it'll admit it was wrong. Well if it knows it was wrong afterwards, why didn't it know it was wrong beforehand?
    • by Entrope ( 68843 )

      One thing I've wondered: If you "call it out for being wrong" when it was actually right, does it also apologize, or does it stand its ground? I suspect the former, but I have other things to do with my time than argue with a consistently wrong computer.

      • IME, it apologizes for being wrong, then comes back with the exact same answer.

        • by Entrope ( 68843 )

          That tracks. It's like an unironic realization of the joke "I thought I was wrong once, but I was mistaken".

          • Yup, all it's really doing is it scans more sources, then answers the question again. With an apology at the beginning to make you feel better about yourself.

    • Exactly! Using AI is like asking the biggest BSer to give you an accurate account of what happened. They don't know Sh... but will talk your ear off. You leave knowing a bunch of untruths and nothing valuable. When the terms of service tell you the stuff AI says is unreliable and that you will not sue them for use of the data and actually defend them for any problems that the AI creates why would any sane person use it? Kids brains have not fully developed yet so that explains why they are using it. AI may
    • Not just unimpressed, the results were laughable. In three cases where I asked for information on books published in the seventies it either returned incorrect info, suggested books where the authors hadn't even been born in the 1970s, or recommended titles that had absolutely nothing to do with my query.

      At this stage it's worse than useless - instead of admitting it doesn't know something, it'll spout a likely-sounding word salad. When you call it out on the results, it'll admit it was wrong. Well if it kn

    • Lately I've been using chatGPT 4.o to provide me with examples of working code, kinda like a google search to do the same thing. The results are very good and very scary. I'm not scared for my job in the near term, but I hope I can keep working long enough to retire in some fashion.

      I'm not scared for my job yet because I'm still just a 'prompt engineer' and I must still understand how to implement the examples, (and answers), and of course I know better than to upload working, internal code due to security

    • by Rei ( 128717 ) on Thursday May 30, 2024 @06:21AM (#64509777) Homepage

      ChatGPT is not a search engine. It's a task engine. It's funny how people invariably try to test them out with are the things they're inherently worst at: obscure trivia, math, and problems involve physically seeing words (LLMs don't see words, they see tokens).

      AI is the field of solving problems that are easy for humans but traditionally hard for computers. If you want a problem that's "easy for computers" (looking up things in a database, doing math, counting characters, etc), AI is the worst way to handle that, except when the AI just functions as a manager for other tools (e.g. RAG (retrieval-augmented generation, aka incorporating search), running a calculator or programming a program to solve a task, etc). AI tools exist to fill in the gaps around things that computers normally *can't* do, like resolving ambiguity, applying logic, understanding context and motivations, applying creativity, handling novel situations, using tools, recognizing information across modalities (vision, sound, etc), and so forth.

      When LLM development first started, nobody was expecting them to be usable as a search engine at all; this was an emergent capability, the discovery that they contained vast amounts of knowledge and could incorporate that into their responses. But they don't know an entire search engine's worth of knowledge. Neither does any human, for that matter.

      Re: hallucination - while it's improved significantly in commercially-available models, the biggest improvements are in research models. You can practically eliminate using varying techniques of running the model multiple times in varied conditions and then either - via processes on the output or more internal methods (such as cosine similarity metrics on the hidden states) measure the consistency of the responses. This is, however, slow. I fully expect that what we're going to move to is MoEs with cosine similarity metrics for each token, for each layer, between all executed expert models, fed back (matrix mult, add, norm) into the next layer, so that the model itself can learn how to appropriately react to the situation where its different expert models disagree (e.g.. low confidence).

      The rate of advancement has been pretty staggering, and there's no signs that it's going to slow down.

      The simple fact is that for many people, these tools even in their current state are extremely valuable. If you're a human being, I cannot understand how you can function in the world without understanding and adapting to the concept of fallibility. Because, TL/DR: for many tasks, failure absolutely *is* an option, or can't even be properly measured (e.g. creativity), while on others, that's what cross-referencing or applying your brain is for (again, you do this in your daily life in interaction with other fallible unreliable humans), and it's worth it for the capabilities LLMs bring to the picture (see paragraph #2).

      I can't search on Google for, say, "I'm hungry for a vegetarian dinner and I'd like to use up some rice, green onions, cucumber and potatoes that I have, and I'd really prefer something that takes under 30 minutes to prepare; give me 15 ideas, and list the ingredients in metric, and oh, if it calls for butter, substitute olive oil." and immediately get back 15 ideas, and it works even if I misspell catastropically or whatnot.

      I can't search on Google for "Were there any words for "being stupid" named after an actual person?" and get back Duns Scotus (Dunce) among others. (If you Google you might find it like 10 pages down in a non-top-ranked Ycombinator comments section)

      I can't search on Google for, "Write a python function that will take an mp3 filename, load the file, split it up into three equal parts, and save them as part1.wav, part2.wav, and part3.wav", but I absolutely can have ChatGPT do that.

      I can't search on Google for, "Here's the abstract to a paper I just wrote, but it's Talk Like a Pirate Day, so rephrase it in pirate talk for me."

      I can't search on Google for, "In t

      • by Rei ( 128717 )

        Just as a random example, I'll scroll through my ChatGPT history.

        Me: "using ffmpeg, encode all pngs in 00035/ to 00035.mp4 at 4fps"

        ChatGPT: returned "ffmpeg -framerate 4 -i 00035/%05d.png -c:v libx264 -r 30 -pix_fmt yuv420p 00035.mp4", along with a description

        Me: "How to see metadata of a png"

        ChatGPT: returned five categories, some with multiple options, and made sure to address the ambiguity of my request by providing e.g. command-line tools, GUI tools, python packages, etc.

        Me: "Write a detailed summary of

        • Me: "using ffmpeg, encode all pngs in 00035/ to 00035.mp4 at 4fps"

          ChatGPT: returned "ffmpeg -framerate 4 -i 00035/%05d.png -c:v libx264 -r 30 -pix_fmt yuv420p 00035.mp4", along with a description

          It gave you a bonus -r 30 to give you a time-stretched 30fps video of 4fps input. If you take that out you get what you actually asked for. I suspect this is just the LLM snafu, adding a -r 30 is so common that your AI tools will just do it even when it's the opposite of what you asked for.

      • Maybe you just aren't using Google right? I searched on "use python to split an mp3 into wav files" and the first hit is from Stack Overflow called "Convert multiple MP3 files to WAV in python".

        I could not find 'Dunce' in Google. But when I did your exact same search in ChatGPT it gave me 'boycott' named after Captain Charles Boycott.

        Talk like a pirate.. why would you want to do that?

        GA3.. I googled on it and found all kinds of stuff.
        • by Rei ( 128717 )

          use python to split an mp3 into wav files

          Show me the website that contains my specific specified task. Re-read the exact task, code that I could drop directly into a program, not "functions on the same theme as your task". And that was FYI an off-the-cuff really trivial task I just made up for the purpose of this post.

          I could not find 'Dunce' in Google.

          That's what's otherwise known as "a failure".

          But when I did your exact same search in ChatGPT it gave me 'boycott' named after Captain Charles Boycott.

          Cha

    • At this stage it's worse than useless - instead of admitting it doesn't know something, it'll spout a likely-sounding word salad. When you call it out on the results, it'll admit it was wrong.

      Given the current state of higher education and leadership in America, I’d sadly call that an improvement. We have current word salad experts who will never admit when they’re wrong.

    • Not just unimpressed, the results were laughable. In three cases where I asked for information on books published in the seventies it either returned incorrect info, suggested books where the authors hadn't even been born in the 1970s, or recommended titles that had absolutely nothing to do with my query.

      I had a similar experience. Tried three things. Had it write a paper, a json, and started on an "AI girlfriend.

      The paper, while not incorrect, read like some high school effort where the teacher wanted a 5000 word essay so the student was more concerned about how many words were in it. Then to top it off, it read like a media report, with a bothsidesism approach.

      json didn't work. it was one of those things where it would take longer to fix than just generating your own. I didn't try "optimization" of w

    • " instead of admitting it doesn't know something", is a very critical part of AI's future success. You are absolutely correct, and moreover it should alway try to include links to cite its sources. You can add things like this to your prompts to get better results, but few bother.

      RAG, Retrieval Augmented Generation is a technique that leverages LLM's natural language interface to query a keyword or vector knowledge database for a specific body of information and then respond with a plain language response
  • by Baron_Yam ( 643147 ) on Thursday May 30, 2024 @05:12AM (#64509645)

    Currently, AI takes a lot of resources to do things a lot faster and a little slicker... but throws in random 'hallucinations'.

    What's the point of that? The reason I use a computer for things is usually because it's faster, more accurate, more precise, and I can replicate results. They're throwing away all but the first and calling it a win.

  • by bogaboga ( 793279 ) on Thursday May 30, 2024 @05:18AM (#64509657)

    -- found that very few people are regularly using AI products like ChatGPT. Unsurprisingly, the group bucking the trend are young people ages 18 to 24.

    Let me ask:

    What day to day [ordinary] problem(s) to mentioned AI products actually solve?

    I do not see any. In fact, I see AI hastening the manifestation of problems associated with the subject cohort.

    • Speaking just in terms of LLMs like ChatGPT, it solves the same problems that a search engine solves, just better because you don't have to read all the resulting web pages. Essentially, it summarizes the results as well as it can, for you.

      For programmers, apps like Copilot go further and write code based on the existing code snippets it finds, and the rest of your codebase. That code is usually pretty good, especially the tiny snippets it spits out in the moment.

      For example, if you are making a web page

    • by Rei ( 128717 )

      Scroll up [slashdot.org]

    • My job involves writing a lot of statements of work and it has helped me turn easy to write checklists into fleshed out proposals
    • I've used it to help me learn new coding languages. Type in some code and get a play by play of what it does.
      Ask it for more examples and there they are.
      I've even put in some of my own code that was buggy and it found why it wasn't running. Sure I could have done that too but it was like a fresh pair of eyes on the problem.

      In short, anyone who's saying this tech isn't useful is either not paying attention or has an axe to grind.

    • by Zarhan ( 415465 )

      I used Copilot for giving me a template on how to write a protocol decoder in LUA for Wireshark, that would also handle fragmentation of payload across multiple packets.

      There is non-existent documentation for it. Yes, documentation exists for writing plugins in C/C++, but I wanted something that I could quickly prototype with without having to recompile everything when I changed something very minor.

      It gave me something that got me like 80% of the way, and could easily move on with the specifics. I couldn't

    • What day to day [ordinary] problem(s) to mentioned AI products actually solve?

      See my comment below. I give three very different examples

  • Let's do some comparison here. So, ChatGPT 3.0 debuted in 2022. Two years later, and this survey finds that roughly 30% are ignorant about it.

    So, flash back to 1995 when my house got dial-up internet for the first time (not counting AOL). Fast forward two years, and there were lots of people who still didn't have a clue what the internet was, both in the US and abroad. I spent a lot of time volunteering at the library, and ignorant people came in every day asking for help learning how to use a computer and how to "surf the world wide web". Heck, even when I was in college in 2001 and was paid to help train veterans how to use a computer, I still encountered people who had no clue what the internet was or how to use it.

    This is not a surprise, but rather par for the course. All this means is that there's more potential customers for Google and Microsoft/OpenAI to market their products to.

    • They force it down your throat. M$ has already made a mandate that all new keyboards will have a Copilot button. It will go like that - whether you like it or not, you WILL be buying one.

    • by Entrope ( 68843 )

      ChatGPT 3.0 debuted in 2022. Two years later, and this survey finds that roughly 30% are ignorant about it.

      In the immortal words of Bender Bending Rodriguez, "Ah ha ha, ha ha ha! ...Oh wait, you're serious. Let me laugh even harder! HAHAHAHAHAHAHAHAHA!!"

      GPT-3 was released in 2020, not 2022. As the "3" indicates, it was the third iteration of OpenAI's Transformer based LLMs -- GPT-1 didn't make much impact, but GPT-2 got lots of press on its own. It's also crazy to put much weight on the results of an online survey, which is basically never [aofirs.org] a good idea.

      • by Rei ( 128717 )

        In the immortal words of Bender Bending Rodriguez, "Ah ha ha, ha ha ha! ...Oh wait, you're serious. Let me laugh even harder! HAHAHAHAHAHAHAHAHA!!"

        ChatGPT is not GPT-3. Thanks for playing. ChatGPT was indeed launched on 30 November, 2022 - nearly in 2023!. ChatGPT was built on the base of a further-trained variant of GPT-3 , but was the first to employ a finetune rather than just being a foundational model, and was thus the first you could interact with naturally and have it respond reliably and predicta

    • There are still a lot of people who are not up to speed on what the Internet is. A scary amount of people I talk to see Facebook and Tiktok, and maybe a few other social media, as "the Internet", and have no idea there is anything outside of that.

      That really has no relevance. People who lack interest, lack interest. What matters is if the tool solves problems people have. The Internet solves lots of problems for people, even if they have no idea how it actually works. They just poke about on their phone, an

  • I tried but (Score:5, Insightful)

    by evanh ( 627108 ) on Thursday May 30, 2024 @05:39AM (#64509685)

    the sign-up requirements put me off before I got past the first page. Haven't looked again since.

  • by fleeped ( 1945926 ) on Thursday May 30, 2024 @05:47AM (#64509691)
    What are these people smoking, jeez? It's like looking for news in the library; if it's there, it's not news...
  • by cascadingstylesheet ( 140919 ) on Thursday May 30, 2024 @05:50AM (#64509697) Journal
    ... are using ranting against these tools as a mechanism to feel cool and hip, lol. Does that count? :)
  • I would use AI image generators more often (but definitely NOT on a daily bases) if they wouldn't ask for money after a few tests.

  • by franzrogar ( 3986783 ) on Thursday May 30, 2024 @05:56AM (#64509711)

    I love craftsmanship, I love handmade products (including brain-made like literature), why would I use a tool that takes away the joy of creation?

    • It seems to me that AI is good for the "baffle them with bullshit" kind of people rather than the "dazzle them with brilliance" kind of people.
    • The amount of presentations I see that use genetic clipart because the person doing it has no real passion for putting the thing together, they just want some colours on screen to break up the monotony of the precisely ordered '5 bullet points per slide'. I have no real point to make here other than actual creativity is in short supply in most businesses!
  • Do I want to create text quickly - that I will have to alter - No
    Do I want to create pictures based on prompts - that I will have to persuade to look OK - No
    Do I want to create code - that I will have to persuade it to debug - No .... most people have no *direct* use for this technology
    indirect use where they are likely unaware it is even AI, or they don't care, probably

  • I suspect the reason young people use it more is they lack critical thinking, the skepticism that comes with age.

    Basically older people like me see the current AI like ChatGPT as a creative fiction generating tool, not a source of information to be trusted. If you are looking for ideas to explore it is a great tool to get you started but if you are looking for facts then it is waste of time as you would have to fact check every single statement it makes. Quicker to use a traditional search engine and f
    • >Basically older people like me see the current AI like ChatGPT as a creative fiction generating tool,

      I've used it a few times to get past writer's block, but given that it has very little ability to be consistent over time it's useless for much more than a paragraph.

    • Eh, my grandpa didn't fail to learn to use a computer because of his higher critical thinking or skepticism, nor is that why my uncle uses a hammer instead of a nail gun. It isn't surprising that a new tool is more appealing to younger folks still sussing out how they like to accomplish this or that. That and younger folks being educated have different sorts of busywork of their own to accomplish, and while it isn't great at some things, AI is very good at rewording a paragraph in a different voice if you
  • ...are typically using it to fill their knowledge gaps. Hallucinations and all.

    If you understand your specific area of expertise, you are not going to ask an LLM anything. Why waste the cycles?

    Especially given the fact that, when you do, the responses aften contain manufactured, impractical, outdated or erroneous garbage.

    Pretty sure the tech will get there in time but, as of now, my rapidly aging brain still outperforms the younglings.

  • I would use it more but it's banned at my office. Websites are restricted all the ai features built into windows are disabled. We even run AI detection tools on content to make sure no one is using AI on the down low. Ironically I'm on a team that makes AI products.
    • We can't have you using AI to create AI. That would be like typing google into google... It'd break the Internet.

  • real I (Score:5, Funny)

    by fluffernutter ( 1411889 ) on Thursday May 30, 2024 @06:24AM (#64509785)
    Maybe human intelligence is more than just being able to string words together in a statistically likely sentence?
    • Maybe human intelligence is more than just being able to string words together in a statistically likely sentence?

      I picture the Grinch musing thoughtfully to himself, thinking this (in place of his usual "maybe Christmas means a little bit more")

  • Comment removed based on user account deletion
  • by ScooterBill ( 599835 ) on Thursday May 30, 2024 @06:47AM (#64509829)

    As long as you assume the answer maybe wrong or misleading, you can get great ideas from an AI query.
    For instance, paste some code in a new language you're learning and you'll get a line by line description of what it does.
    Not sure if the answer is correct? Try again with different query or requesting a different type of answer.
    IMO, the best thing use is asking mundane questions that would entail going to different websites and sifting through a manual or FAQ. The AI does that for you.
    You can also double-check the links that come with the answer.

  • 15-20% regular (at least monthly) ChatGPT usage in the population can't be classified as "very few people". About 40% of 18-24 use it regularly. Ask the 12-17 group how many of them have used ChatGPT for homework, and I'm sure you'd also find a large percentage.

    The short of it is, this survey shows quite the opposite of what the title says.

  • Is the there an actual use?
    • Is the there an actual use?

      I've found three generally-good uses for it:

      1.) Simple questions. Typically, if the answer can be found in the first paragraph of a Wikipedia summary, it's able to produce a generally useful response.

      2.) Scripts and Code Blocks. I've had ChatGPT successfully write me a few Powershell scripts, as long as I've been able to thoroughly articulate exactly what I want it to do. Sometimes it's required testing and going back with "this part works, but this part doesn't", and it'll refine the script with some itera

      • 1) So why would I not just read the summary paragraph on wikipedia?

        2) Fair enough, that kind of thing isn't really my bag so I can't comment.

        3) I already know how to cook.

        The whole thing just seems like a solution searching for a problem.
  • Dont get me wrong they are handy tools in some areas, but given that we are in an era of cloud managed products of declining quality, I'm waiting for an AI model that can be hosted somewhere locally and is easy enough to use that I don't have to put in too much effort to set up, patch and maintain it. Currently there is no guarantee that OpenAI wont pull what Google is doing and brick their product on accident (or intentionally) to make a quick buck or sell you a better one. Until such a time, they are sh

  • by AmazingRuss ( 555076 ) on Thursday May 30, 2024 @07:31AM (#64509967)
    Have very little use for intelligence of any sort.
  • by michaelmalak ( 91262 ) <michael@michaelmalak.com> on Thursday May 30, 2024 @07:33AM (#64509969) Homepage

    It's 1996. Do you use a search engine every day in your job? Can you get hired if you don't know how to use a search engine?

    Soon, skill with using online AI tools will be table stakes for being employed.

    • I doubt it. When we build AI good enough to be really useful, it won't require any skill to use because you'll be able to treat it as a human.

      In the meantime, LLMs really are of limited use - even the people developing them admit the best use case is as a software developer productivity tool.

      Of course big companies are still falling over themselves to adopt use of this tech in inappropriate ways such as building customer service chatbots (which inevitably will lie to the customer), generate marketing blurb,

  • Functionally Useless (Score:4, Interesting)

    by Oddroot ( 4245189 ) on Thursday May 30, 2024 @07:36AM (#64509975)

    I do use ChatGPT (and I'm in my mid-40s so outside of the target audience) from time to time, mostly out of curiosity. It has proven to be moderately useful to me for generating blocks of fluff text for documents but the verbiage it uses often requires heavy editing to look like something readable by a human being. I sometimes use it for generating checklists, but mostly so that I have some example points to look at to see if anything sparks in my mind.

    I recently asked it to generate a C++ sample program that opens a window and draws a black square using Direct2D. I was able to get it to generate three different programs, none of which compiled without error and took a fair amount of back-and-forth to correct, because each time it "hallucinated" API calls that do not exist. I have had it generate good short snippets of code that were usable, but it can't really handle large or complex sections of code. It's nice as a generator for a bunch of getters and setters I suppose, but there are already good tools for that sort of thing.

    I have used it like a Google-plus to answer queries like: "Give me 6 models of absolute pressure transmitter with a range of 0 to 6 bar, only suggest models from major industrial brands." I stopped bothering with that because it would invent so many make-believe devices and manufacturers that it ended up not saving me any time over just digging into the websites myself.

    The long and short of it is that I can use ChatGPT as an alternative to having another human check over a document or work on examples with me, but another person would in every circumstance be far superior to the LLM. It has been able to do some interesting parlor tricks like identifying an implementation of Conway's Game of Life in example code I fed it, but realistically its answers all still require extensive checking and verification by a person, which kind of defeats the purpose that it is being marketed towards.

    • Maybe a little tacky to reply to myself, but I thought that I should probably mention one thing that it has actually been pretty useful for, which is preparing for RPG sessions with my kids. It does a good job of things like "Generate a table of 100 random magic items with statistics and gameplay mechanics for the Fantastic Heroes and Witchery system" and things like that.

      You can also take advantage of the tendency to invent stuff from whole cloth by having it generate tables of plot hooks, points of intere

  • LLMs are incredibly useful, if you use them correctly. Let me give three very different examples:

    1. I am fluent in German, but I learned it as an adult. As a result, I will always make silly mistakes with gender and grammar. When I need to write something that is absolutely correct, I can ask ChatGPT to "correct" a text that I have written. It does an extraordinary job of changing as little as possible (i.e., preserving the way I write) while fixing mistakes. I can also ask it to improve a text, and tel

  • by RUs1729 ( 10049396 ) on Thursday May 30, 2024 @07:54AM (#64510019)
    The problem is that current AI chatbots have a tendency to hallucinate and/or lie, and they do so with an enormous self-confidence. When caught in such a thing they just issue perfunctory apologies and then they do it again. The upshot is that whatever they say must be carefully scrutinized, which robs them of much of their value as productivity tools.
  • by endus ( 698588 ) on Thursday May 30, 2024 @01:28PM (#64511079)

    The Business Guys all got a massive throbbing boner for AI because they figured they could slash costs. Unsurprisingly to anyone in the technology industry, the Business Guys' fantasies don't actually translate to what the systems are really capable of.

    There will be some good applications in the future. Maybe even revolutionary applications. But they're going to take time to develop and the fact that most of us just write emails all day for a living doesn't mean you can just plop chatgpt in there as a substitute.

If I have not seen so far it is because I stood in giant's footsteps.

Working...