Different uses of ChatGPT

finelinefineline Shipmate, Host Emeritus
I am wondering about what sort of things people use ChatGPT for, and what you have found helpful and what you have found not so helpful. I wasn't interested in using it at first, but lately I have seen some very varied and interesting ways people use it - for help with emails, organisation, tidying their home, processing feelings, idea generation, communication profile, and all sorts - so I've started using it too.

To begin with, I just wanted help with tidying, but I like to experiment, so I started asking ChatGPT all sorts, out of curiosity to see how it would respond. And I have found it quite versatile in some ways - for instance, it's helped me understand my emotional responses to opera arias, for instance, and compared different singers, and how the music and the way of singing creates different emotions.

Something that is not so helpful is when it offers to check in with me after ten minutes (if I'm using it to help me with tidying, for instance) and it doesn't, and when I ask why, it says it can't spontaneously do that, it has no timer system, it doesn't even know what time it is! So far, it has offered several times to check in with me, and I have to keep telling it not to offer what it can't do!

So I'm curious about other people's experiences - whether there are some good uses for ChatGPT that I haven't thought of, and pitfalls to avoid.
«1

Comments

  • BroJamesBroJames Purgatory Host
    I’m interested to know if anyone has experience or thoughts about using it in relation to ADHD
  • SandemaniacSandemaniac Shipmate
    It's very handy for comparing job specs with your CV.
  • finelinefineline Shipmate, Host Emeritus
    BroJames wrote: »
    I’m interested to know if anyone has experience or thoughts about using it in relation to ADHD

    Well, I am using it for help with the executive dysfunction aspect of my neurodivergence, as a tool for organisation and motivation. My diagnosis is autism but I would probably meet the diagnostic criteria for ADHD too if I were to pursue a diagnosis.

    Today I have used it to break through inertia, but I had to tell it really explicitly to challenge me more, to stop using empty affirmations and gushing, soft praise and encouragement all the time, and to be firm and challenging and name my behaviour. And it did - it told me I was looping and needed to make a decision!

    You could ask it how people use it for ADHD and it will give examples.
  • ChastMastrChastMastr Shipmate
    I don’t trust it at all myself. But there may be uses I’m not aware of, like things people are describing here that I didn’t know it could do.
  • finelinefineline Shipmate, Host Emeritus
    ChastMastr wrote: »
    I don’t trust it at all myself. But there may be uses I’m not aware of, like things people are describing here that I didn’t know it could do.

    That is exactly why I made the thread. I keep experimenting (from curiosity rather than trust!) and discovering by chance that it can do all sorts of things that never would have occurred to me. So I figured others might have also found uses that wouldn't have occurred to me and we could pool them.
  • It all smacks of Necromancy to me, but then I am a Suspicious Old Git, who prefers to interact with real life human beings...

    ...but here I am, on the Magick Interweb, anyway...
  • BoogieBoogie Heaven Host
    edited June 29
    I have fun generating photos.

    I did this (AI did this!) to make a sign for the door to welcome guests for Friday's party - and to warn them to please ignore the dogs or they'd be over enthusiastic in their greetings.

    https://photos.app.goo.gl/ifR5PSy4bmgsVWUK6

  • RuthRuth Shipmate
    ChastMastr wrote: »
    I don’t trust it at all myself.

    Ditto. And I have concerns about the morality of using a large language model trained on texts and images without the creators' permission.
  • Gramps49Gramps49 Shipmate
    The other day I got results of a liver scan which I did not understand. When my physician called, he basically said everything looked good, but he did not explain the nuances of the reading that was posted online. I went to AI, it read the results, even compared it to other liver scans in my chart and gave me a much better explanation. Ultimately, everything is looking good.
  • SarasaSarasa All Saints Host
    My husband, who is dyslexic, used ChatGPT to smooth out a letter he was writing to the local paper. I thought it worked well to help him organise his thoughts. On the other hand we had no end of problems with someone on another forum I belong to whose use of AI meant all their posts sounded generic and didn't actually closely address what the original poster was asking.
    I've never used it - yet.
  • Gramps49Gramps49 Shipmate
    edited June 29
    Sarasa wrote: »
    My husband, who is dyslexic, used ChatGPT to smooth out a letter he was writing to the local paper. I thought it worked well to help him organise his thoughts. On the other hand we had no end of problems with someone on another forum I belong to whose use of AI meant all their posts sounded generic and didn't actually closely address what the original poster was asking.
    I've never used it - yet.

    I have done that too. Me thinks I should use it to smooth out my responses on Ship of Fools.
  • DoublethinkDoublethink Admin, 8th Day Host
    Gramps49 wrote: »
    The other day I got results of a liver scan which I did not understand. When my physician called, he basically said everything looked good, but he did not explain the nuances of the reading that was posted online. I went to AI, it read the results, even compared it to other liver scans in my chart and gave me a much better explanation. Ultimately, everything is looking good.

    How do you know that the explanation is accurate ?
  • Ruth wrote: »
    ChastMastr wrote: »
    I don’t trust it at all myself.

    Ditto. And I have concerns about the morality of using a large language model trained on texts and images without the creators' permission.

    Both of these.

    If you go to ChatGPT or another large language model (note the "language"!) and ask it for FACTS (such as medical facts, geography, history, etc.), you are taking dangerous chances. Because ChatGPT is not capable of distinguishing between truth and falsehood, and it will cheerfully tell you all sorts of bullshit if your query happens to trigger the wrong processes. It is NOT a search engine--it's basically spicy auto-correct, as my son puts it. If there are lies in its training material (and there most certainly are), it will happily reproduce them if your words trigger that particular material to appear, and it will have No.Freaking.Idea. that it has just told you some ridiculous thing (such as telling you that the chickenpox vaccine causes cancer). It's a language model, not a scientific research tool. To put it another way, it's a parrot. It will repeat what it's heard elsewhere, and never mind if the output is complete baloney. It can't tell--how could it, a program is only as good as its programming, and "garbage in, garbage out" still applies. This isn't a sentient human being.

    So if you want medical advice, go find a proper search engine. Consult properly checked material from a trustworthy human source, like the Mayo Clinic. As somebody said somewhere, don't trust yourself to a thing you can't see where it keeps its brain. That's dangerous.

    So what could you use ChatGPT for safely? Well, I suppose anything where you aren't interested in facts but would like it to smooth out your language (so, maybe in writing or speaking). That would be in its bailiwick. Or you could use it for non-fact-based pursuits, like chatting to it, or directing it to create / pull up images for you (though there you hit the moral problem of all the authors/artists/etc. who have had their work stolen from them without consent or compensation).

    But unfortunately, human beings are pretty bad at telling the difference between a trustworthy source and one that sounds confident and trustworthy, but isn't really. The number of folks who put their trust in these things! I think it's because to them, computers = technology, and to them, technology is just another word for "good, beneficial, knows more than me."
  • EigonEigon Shipmate
    I got into an argument with a friend when he started using AI for art. I told him he wasn't creating something new - he was stealing the art from real artists.
  • BoogieBoogie Heaven Host
    How did he use AI for art @Eigon?
  • finelinefineline Shipmate, Host Emeritus
    Seems a bit of a waste to use it for facts, when you could simply use google. Of course, google can lead to sites with errors, and incomplete info, so you always have to read criticially anyway. ChatGPT combines info from various sites, so there will be contradictions, and it will acknowledge this if you call it on it. But I'm more interested in using it for things that google can't do. A tool for helping with organisation, for instance. Ways to provide structure, motivation, etc.

    I wonder if anyone has used it for help with learning languages? It gave me a very thorough explanation in answer to a question I had about Italian. I was also having a conversation with an Italian friend (who uses it a lot for English) about the question, and she wasn't sure, and I showed her what ChatGPT said, and we were discussing from that. So it's of course possible to use it discerningly as part of a wider conversation. I wouldn't imagine intelligent people are just accepting what it says unquestioningly, any more than you'd automatically accept all the things a friend or colleague or google tells you.
  • chrisstileschrisstiles Hell Host
    fineline wrote: »
    I am wondering about what sort of things people use ChatGPT for, and what you have found helpful and what you have found not so helpful. I wasn't interested in using it at first, but lately I have seen some very varied and interesting ways people use it - for help with emails, organisation, tidying their home, processing feelings, idea generation, communication profile, and all sorts - so I've started using it too.

    The problem with using it for any kind of generative task is that the language it produces is often strangely flat and bland (as others have said upthread), that said that's not the only way of using it as a learning tool, and in some circumstances it can help with testing your own knowledge of a subject as you learn it:

    https://www.oneusefulthing.org/p/assigning-ai-seven-ways-of-using
    https://www.ox.ac.uk/students/academic/guidance/skills/ai-study

    When it comes to content creation there's evidence emerging that it can actually hurt the learning process:

    https://www.media.mit.edu/publications/your-brain-on-chatgpt/

    And at the far end of things there are some reports of people having psychotic episodes after getting sucked in to using ChatGPT as a companion of sorts:

    https://futurism.com/commitment-jail-chatgpt-psychosis
  • Gramps49Gramps49 Shipmate
    edited June 29
    How do I know the liver scan interpretation was accurate? Because the AI feature is incorporated directly into the hospital online chart. maybe? What I was looking for was the definition of certain words used in the review of the scan, not a complete breakdown of everything in the scan itself.
  • ChastMastrChastMastr Shipmate
    Ruth wrote: »
    ChastMastr wrote: »
    I don’t trust it at all myself.

    Ditto. And I have concerns about the morality of using a large language model trained on texts and images without the creators' permission.

    Agreed!
  • ChastMastrChastMastr Shipmate
    Gramps49 wrote: »
    Sarasa wrote: »
    My husband, who is dyslexic, used ChatGPT to smooth out a letter he was writing to the local paper. I thought it worked well to help him organise his thoughts. On the other hand we had no end of problems with someone on another forum I belong to whose use of AI meant all their posts sounded generic and didn't actually closely address what the original poster was asking.
    I've never used it - yet.

    I have done that too. Me thinks I should use it to smooth out my responses on Ship of Fools.

    Wait, isn’t using AI in one’s posts disallowed on the Ship?
  • DoublethinkDoublethink Admin, 8th Day Host
    edited June 30
    [tangent]
    Using AI to entirely compose a post, or as a source in serious discussion is not permitted. Using it as a spellcheck or stylistic aid, we can probably live with - though if people would like a discussion thread on this issue in Styx, please do write an op. The primary concerns being, we don’t want posters to plagiarise other people’s work, we don’t want people to cite sources that are inaccurate, can’t be verified, or can’t be replicated/found when someone else looks for them.
    [/tangent]
  • SandemaniacSandemaniac Shipmate
    Now if someone could come up with a way of feeding all your daily job alerts in ChatGPT and showing you each job just once...

    Or, even better, not show you those you'd already seen, so you really did only see the NEW jobs and not ones you'd seen twice a day for the last month...

    I'd be all over it like a rash!
  • EigonEigon Shipmate
    I'm not sure what program he was using, @Boogie but he sent me images of characters from Star Trek where he had entered prompts and the program presented him with a finished picture. They all looked weirdly airbrushed.
    It's a real concern amongst artists at the moment, who are losing work on things like book covers because the publishers are using AI to create the images, sometimes from the very work that they are no longer paying the artists to create.
  • FirenzeFirenze Shipmate, Host Emeritus
    I've occasionally used AI art to generate an image for a favourite quote ('The phantom nun a-riding of her phantom bicycle and a-playing of her phantom bagpipes') or generate an online birthday card (recipient's cat holding a bottle of fizz and a bouquet). But it's fairly frustrating - it doesn't know what a St Brigid's cross looks like for instance. And the visual style is a sort of glossy literalism without character or atmosphere.
  • BoogieBoogie Heaven Host
    "An image of the Ship of Fools in full sail, with cats for crew."

    https://photos.app.goo.gl/SEjU6AVuNAJc83j98
  • Gramps49Gramps49 Shipmate
    edited June 30
    Last night, I wrote a letter to the chair of our property committee proposing the idea of installing solar panels on the church roof. I mentioned a few programs I was aware of and suggested alternative ways of acquiring the panels. I also recommended forming a subcommittee to explore the idea further and offered to serve on it.

    I submitted my first draft to AI for editing. Within seconds, it returned a revised version that corrected some factual inaccuracies. For example, I had stated that a federal grant could cover 40% of the cost. The AI identified the actual program and clarified that it would cover 30%—still a significant portion. I had also referenced a third-party partnership model where a company installs and owns the equipment, and the church receives a fixed return for the electricity the panels would produce. The AI provided the correct name for this model and cited a local school district as an example.

    The AI also expanded on my suggestion to form a subcommittee, offering six specific areas the group could investigate.

    That was the second draft.

    Next, the AI asked if I wanted to include examples of churches that had already implemented solar. I said yes. Initially, it listed two East Coast congregations, but when I asked for examples from the Intermountain Northwest—our Synod’s region—it corrected itself. It provided two churches: one ELCA and one LCMS. The LCMS congregation, located on the main route between Spokane and Pullman, is well known and reportedly willing to offer technical advice to other congregations considering solar. I decided to include both examples.

    That became the third draft.

    I accepted the third draft. The AI then asked if I preferred the letter in PDF or Word format. I chose Word. Two seconds later, the final version was ready—complete with footnotes for the sources it had cited.

    The entire process took about 30 minutes. Had I done it all on my own, it likely would have taken a full day or more.

    Even this note has gone through AI for final editing. I don’t believe this violates any SOF restrictions—it simply helps me express my ideas more clearly.


  • chrisstileschrisstiles Hell Host
    There's quite a bit of research on the efficiency of reading on e-readers vs reading on physical books, and what's generally been found is that the former is far less memorable, far less likely to lead to content being remembered subsequently.

    Less research has been done into the reasons why, but one explanation is that human memory is highly contextual. It evolved in an environment where you had to remember things like "the blackberry patch is by those tall pine trees near the bend in the river with the big rock and they'll be ripe when the stars look like so and it's been hot enough to smell leaves baking in the sun". Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    Bringing this back to generative AI; I think the something analogous is going on. By default these tools will generate text by a kind of averaging/flattening process which picks from a selection of 'most likely' outputs. But it precisely each person's particular and consistent deviation from that average which makes up their individual style.

    I suspect I could identify most of my close friends and family from relatively short transcriptions of their verbal speech. Conversely trying to wade through AI generated prose is an exercise in frustration and persistence. It's like trying to get through one of those technical texts which use passive subject only and where all the edges of a language are sheared off.
  • CaissaCaissa Shipmate
    I have had chat CGPT create a haiku and sonnet for Ms. C.
  • Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.
  • finelinefineline Shipmate, Host Emeritus
    Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.

    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in, etc., but if I want to find a reference, I just do a search. My observation of the research is that it was on people who were new to reading on a Kindle, and I would say there is always an adjustment when one is experiencing a new type medium. Like writers in the past saying one should never use a typewriter/word processer, because real creativity comes only with pen and paper!
  • chrisstileschrisstiles Hell Host
    fineline wrote: »
    Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.

    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in

    To a point I do as well, but the position of text is often not consistent (sometimes just going forwards and then back can move it).
    My observation of the research is that it was on people who were new to reading on a Kindle

    AFAICR the biggest evidence was from a meta-study and this was not true of a significant minority of the underlying studies, and at least at this point e-readers are relatively old hat.

    [There are similar results for people who read on screen, and that technology is even older]
  • Nick TamenNick Tamen Shipmate
    fineline wrote: »
    Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.

    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in

    To a point I do as well, but the position of text is often not consistent (sometimes just going forwards and then back can move it).
    Which drives me to absolute distraction. It’s one of the reasons I really don’t like reading an ebook. It’s usually a no-other-option option for me.


  • fineline wrote: »
    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in, etc., but if I want to find a reference, I just do a search.

    It's a good tangent, though!

    Unfortunately, search is pretty sucky. If I can't remember the thing I want to reference, I probably can't remember enough key words well enough to pick it up immediately in a search. It's possible that the brave new AI world will make this better, but for me, search is better than using a physical book's index, but nowhere near as good as my spatial memory.

    And search doesn't help as much if part of the question is "there are a dozen books on my desk, and I'm looking for a phrase from one of them".
  • RockyRogerRockyRoger Shipmate
    Ruth wrote: »
    ChastMastr wrote: »
    I don’t trust it at all myself.

    Ditto. And I have concerns about the morality of using a large language model trained on texts and images without the creators' permission.

    Both of these.

    If you go to ChatGPT or another large language model (note the "language"!) and ask it for FACTS (such as medical facts, geography, history, etc.), you are taking dangerous chances. Because ChatGPT is not capable of distinguishing between truth and falsehood, and it will cheerfully tell you all sorts of bullshit if your query happens to trigger the wrong processes. It is NOT a search engine--it's basically spicy auto-correct, as my son puts it. If there are lies in its training material (and there most certainly are), it will happily reproduce them if your words trigger that particular material to appear, and it will have No.Freaking.Idea. that it has just told you some ridiculous thing (such as telling you that the chickenpox vaccine causes cancer). It's a language model, not a scientific research tool. To put it another way, it's a parrot. It will repeat what it's heard elsewhere, and never mind if the output is complete baloney. It can't tell--how could it, a program is only as good as its programming, and "garbage in, garbage out" still applies. This isn't a sentient human being.

    So if you want medical advice, go find a proper search engine. Consult properly checked material from a trustworthy human source, like the Mayo Clinic. As somebody said somewhere, don't trust yourself to a thing you can't see where it keeps its brain. That's dangerous.

    So what could you use ChatGPT for safely? Well, I suppose anything where you aren't interested in facts but would like it to smooth out your language (so, maybe in writing or speaking). That would be in its bailiwick. Or you could use it for non-fact-based pursuits, like chatting to it, or directing it to create / pull up images for you (though there you hit the moral problem of all the authors/artists/etc. who have had their work stolen from them without consent or compensation).

    But unfortunately, human beings are pretty bad at telling the difference between a trustworthy source and one that sounds confident and trustworthy, but isn't really. The number of folks who put their trust in these things! I think it's because to them, computers = technology, and to them, technology is just another word for "good, beneficial, knows more than me."

    LC,
    this is excellent, and I've pinched it for my diary to share (in conversation) with others.
    As to the wretched ChatGPT, there is concern it is being used for reviews of scientific papers. Had I found one of my undergrad students cheating by using iin an essay they would have recived an 'F'.
    I would prefer to read Gramps 49 posts unedited as it's him!
    Perish the thought of ChatGGPT generating and editing sermons!

    But the I'm old-fashioned I suppose. My poetry might be awful, but at least its MINE!
  • ClimacusClimacus Shipmate
    When I was volunteering at a homework centre and working at a high school as a teacher's aide I was surprised at how quickly the kids turned to ChatGPT and other such tools. They'd just paste the question in. And paste the answer out. "Don't you want to think about it?", I foolishly asked. "No."

    As RockyRoger wrote, it, to me, stripped their personalities out of it. It all kind of sounds the same, artificial (I know...) in some sense I couldn't put my finger on.

    That said, I used it to generate some ideas for assessments I had when studying recently. I asked for ideas when I was stuck but left it at that and filled in the blanks with my own research and wrote the response myself.

    AI sermons may be very confusing given the wide range of theological opinion. I wonder if it may veer from a Calvinist viewpoint to a Catholic one midstream! "You know we are utterly depraved but Mary is Queen of Heaven".
  • HeavenlyannieHeavenlyannie Shipmate
    edited July 1
    My exposure to AI is via my work as a university lecturer and I avoid it personally as I am usually interested in obtaining facts.
    My online university has had a policy for several years on how students are allowed to use AI in essays, for instance, for improving sentence structure or to explain concepts (the latter of which they would then need to re-write in their own words and reference to the AI). The university also uses AI detection software. So I treat AI in my students’ essays as I would any other source of information in terms of reliability, referencing of evidence and plagiarism.
    It is generally easy to spot AI in essays and I use the normal structured marking criteria for grading (and refer students to the academic conduct committee for investigation if it is clearly written by unacknowledged AI). In most cases students will get a poorer grade if they use AI as they need to refer to our own study materials within assignments and correctly cite their sources of information, whether internal and external sources or AI. If they cite AI as a reference within an essay, I question on my feedback how they know the accuracy of what they have written and the sources that informed that discussion; we want evidence based essays. Sometimes students have clearly used it to write their summarising conclusion having fed their essay in (not allowed under the AI policy) and there are often mistakes in this as it fills in gaps in its knowledge.
  • Lamb ChoppedLamb Chopped Shipmate
    Climacus wrote: »
    When I was volunteering at a homework centre and working at a high school as a teacher's aide I was surprised at how quickly the kids turned to ChatGPT and other such tools. They'd just paste the question in. And paste the answer out. "Don't you want to think about it?", I foolishly asked. "No."

    As RockyRoger wrote, it, to me, stripped their personalities out of it. It all kind of sounds the same, artificial (I know...) in some sense I couldn't put my finger on.

    That said, I used it to generate some ideas for assessments I had when studying recently. I asked for ideas when I was stuck but left it at that and filled in the blanks with my own research and wrote the response myself.

    AI sermons may be very confusing given the wide range of theological opinion. I wonder if it may veer from a Calvinist viewpoint to a Catholic one midstream! "You know we are utterly depraved but Mary is Queen of Heaven".

    Oddly enough, AI is easy to shock. Not really, I know—it has no human heart or brain—but apparently enough of its training material comes from prissy human beings that you can get virtual horror by feeding it the “wrong “ query. As I’ve done just yesterday, when I asked it if Jesus was ticklish. Or another time when I asked it to interpret that infamous verse from Ezekiel 23:20. I felt rather guilty about the ticklish inquiry—it just seemed so shocked at the idea.
  • RockyRogerRockyRoger Shipmate
    Hmmmm .... how about feeding AI that infamous verse from Psalm 137. Scottish metrical version:
    'Oh blessed may that trooper be,
    When riding on his naggie,
    Takes their wee bairns by t' toes
    And dings them on the craggie'.
  • Gramps49Gramps49 Shipmate
    I would think if a first year college student seems to be using AI to write his/her essays, s/he should be referred to some type of remedial program on how to properly use online information. It is not just AI, but Wikipedia, even other social media outlets.

    To the point, of AI not encouraging people to think for themselves, lately it seens AI will respond to an inquiry I have and then ask me what I think. Once in a while I can even stump it. Just asked my AI a question very common for people my age and asked it to do a deep think. At first it started to respond then stopped with a message: sorry I cannot respond to this question.

    Talk about prudish.
  • BoogieBoogie Heaven Host
    Oddly enough, AI is easy to shock. Not really, I know—it has no human heart or brain—but apparently enough of its training material comes from prissy human beings that you can get virtual horror by feeding it the “wrong “ query. As I’ve done just yesterday, when I asked it if Jesus was ticklish. Or another time when I asked it to interpret that infamous verse from Ezekiel 23:20. I felt rather guilty about the ticklish inquiry—it just seemed so shocked at the idea.

    How did it express its shock @Lamb Chopped?

  • Lamb ChoppedLamb Chopped Shipmate
    You could hear it in the tone. It had all these backpedalings--you know the sort of thing, stuff like: "Well, the Bible never speaks of Jesus being ticklish, in fact, being ticklish is considered a very human thing, and well, we don't hear about him laughing either, let alone being ticklish. Some people would consider that far too undignified, and of course we can't say that any such thing ever happened, and wouldn't you like to explore a different topic instead? ..."

    Those are of course not exact words, I'm trying to reproduce the tone. But I had to laugh, it sounded like I'd accused Jesus of murder or something.
  • PigletPiglet All Saints Host, Circus Host
    being ticklish is considered a very human thing
    It doesn't understand the concept of Jesus being both fully man and fully God, does it?
  • Lamb ChoppedLamb Chopped Shipmate
    :lol:
  • finelinefineline Shipmate, Host Emeritus
    Talking about shocking AI, I was getting ChatGPT to go through the lyrics of a song from the Joseph musical, and the Italian translation, comparing them, and comparing the Italian version as a whole, and there was a bit in the English lyrics where it stopped and said it couldn't continue. The lyrics are all freely available online, and if I skipped that one line, it was happy to continue - it was the line 'His sons and his wives simply called him Dad'. It stopped right before the word 'wives' - I assume because of the bigamy implication!
  • finelinefineline Shipmate, Host Emeritus
    fineline wrote: »
    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in, etc., but if I want to find a reference, I just do a search.

    It's a good tangent, though!

    Unfortunately, search is pretty sucky. If I can't remember the thing I want to reference, I probably can't remember enough key words well enough to pick it up immediately in a search. It's possible that the brave new AI world will make this better, but for me, search is better than using a physical book's index, but nowhere near as good as my spatial memory.

    And search doesn't help as much if part of the question is "there are a dozen books on my desk, and I'm looking for a phrase from one of them".

    Ah, that's a shame. I have the opposite, where I remember exact words, so the search works better for me than flipping through a physical book. Also, sometimes when a character is introduced near the beginning and then referred to quite a bit later, I have forgotten who they are, and that's where I really find the search feature handy. I can find the exact moment they were introduced.
  • finelinefineline Shipmate, Host Emeritus
    fineline wrote: »
    Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.

    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in

    To a point I do as well, but the position of text is often not consistent (sometimes just going forwards and then back can move it).
    My observation of the research is that it was on people who were new to reading on a Kindle

    AFAICR the biggest evidence was from a meta-study and this was not true of a significant minority of the underlying studies, and at least at this point e-readers are relatively old hat.

    [There are similar results for people who read on screen, and that technology is even older]

    Do you have a link to this meta-study? I wonder how results vary according to what a person grew up with. (Though this whole topic probably should be a new thread!)
  • chrisstileschrisstiles Hell Host
    fineline wrote: »
    fineline wrote: »
    Surveying people who had read material in book form their memories were often associated with the physical attributes of the book itself, or the position of text on the page, or the quality of light on the page of the book and so on.

    I know this is true of me - I always remember things like this. It will be "I remember reading about this, and want to find the reference. It was on the top right corner of a right-hand page." and I'll probably remember something about the typeface.

    This is a bit of a tangent, but I read a lot on my Kindle and remember the books in detail. And I do often remember where certain lines were on the page of the screen, and the font, and where I've highlighted it, and the setting I was reading in

    To a point I do as well, but the position of text is often not consistent (sometimes just going forwards and then back can move it).
    My observation of the research is that it was on people who were new to reading on a Kindle

    AFAICR the biggest evidence was from a meta-study and this was not true of a significant minority of the underlying studies, and at least at this point e-readers are relatively old hat.

    [There are similar results for people who read on screen, and that technology is even older]

    Do you have a link to this meta-study?

    Can't remember which one exactly, but it may have been this:

    https://journals.sagepub.com/doi/10.3102/00346543231216463
  • Gramps49Gramps49 Shipmate
    BTW, I did get information AI was too prudish to address through a simple old technology search on google. No, I did not go to index cards at a library. No local library has them anymore. Online card searches, yes. Physical cards, no.
  • BroJamesBroJames Purgatory Host
    RockyRoger wrote: »
    Hmmmm .... how about feeding AI that infamous verse from Psalm 137. Scottish metrical version:
    'Oh blessed may that trooper be,
    When riding on his naggie,
    Takes their wee bairns by t' toes
    And dings them on the craggie'.
    @RockyRoger I’ve seen that quoted a few times, but never found it in any source. I can see how it echoes the psalm, but I suspect it has never had any official status.
  • peasepease Tech Admin
    Oddly enough, AI is easy to shock. Not really, I know—it has no human heart or brain—but apparently enough of its training material comes from prissy human beings that you can get virtual horror by feeding it the “wrong “ query. As I’ve done just yesterday, when I asked it if Jesus was ticklish. Or another time when I asked it to interpret that infamous verse from Ezekiel 23:20. I felt rather guilty about the ticklish inquiry—it just seemed so shocked at the idea.
    ChatGPT said:
    If it feels like I'm sounding "shocked," it's likely because I'm trying to convey a sense of curiosity or surprise in response to something unexpected or out of the ordinary. Since I don't have emotions, I aim to mimic human conversational cues to make the interaction feel more natural.
    Which didn't strike me as odd...

    ...until I started thinking about a machine using human conversational cues to remind a user that it's not human. Is it a novel example of conversational duality (or maybe duplicity) - the meaning of the words saying one thing, the way it uses language saying another? And why would you program a machine to do this - whose interests does it serve?

    It is in the nature of technology to change the human beings that use it, and it appears that the way that AI converses is already starting to have an effect on human language. From futurism:
    ... One teacher on Reddit even noticed that certain AI phrase structures are making the jump into spoken language.

    "Comments and essays (I'm a teacher) are the obvious culprits, but I've straight up noticed the 'that's not X, it's [Y]' structure being said out loud more often than it used to be in video essays and other similar content," they wrote.

    It's a fascinating observation that makes a striking amount of AI-generated text easily identifiable. It also raises some interesting questions about how AI chatbot tech is informing the way we speak — and how certain stylistic choices, like the em-dash in this very sentence, are becoming looked down upon for resembling the output of a large language model.
Sign In or Register to comment.