Kevlin HenneyDec 6, 2022
As the term 'intelligence' is used in 'artificial intelligence', it is perhaps worth asking, what kind of intelligence do AI systems represent? Confident and detail focused, but naïve, easily misled and lacking in common sense, worldliness, situational awareness, fairness, imagination, empathy, emotional intelligence and social intelligence 🤔
Patrick M. HausenDec 8, 2022
Show threadAndrew Feeney

@kevlin

Christine Lemmer-Webber (@cwebber) described ChatGPT as Mansplaining As A Service, and honestly I can’t think of a better description. A service that instantly generates vaguely plausible sounding yet totally fabricated and baseless lectures in an instant with unflagging confidence in its own correctness on any topic, without concern, regard or even awareness of the level of expertise of its audience.

Dec 6, 2022 at 9:54amWeb
Show threadPatriziaDec 6, 2022
@andrewfeeney @kevlin @cwebber At least it only does so when *asked*!
Show threadKevlin HenneyDec 6, 2022
@patrizia @andrewfeeney @[email protected] That may be a matter of timing than of essence: just wait for the spam and the bots... 😨
Show threadYaoi Box BalloonJan 10, 2023
@kevlin @patrizia @andrewfeeney @[email protected] Wait so are we seeing a market for a bot blocker..? 🤓
Show threadPeter VerhasDec 7, 2022

@patrizia @andrewfeeney @kevlin @cwebber no, Patricia, no! Let me correct you!

Just kidding, also deliberate mispell for the sake of demonstration. Sorry, hope you get it.

Also: with this post I passed the “I am not a robot” captcha test.

Show threadSibrosanJan 12, 2023

@verhas @patrizia @andrewfeeney @kevlin @[email protected]

"deliberate mispell"

😜

Show threadJust B 🇨🇦​Dec 7, 2022
@patrizia @andrewfeeney @kevlin @cwebber hahaha so true! 🔥👏
Show threadJake MannixDec 7, 2022

@patrizia @andrewfeeney @kevlin @cwebber and has no ego if you tell it it's wrong, and it can't fire you.

I kinda love it. It's like, we all hate having someone yell at us and boss us around, but we also *pay* personal trainers to do exactly that.

As long as we're in control, and we decide whether to follow the advice, #chatGPT is a fantastic tool.

I *do* worry about having its outputs emitted willy-nilly into the world as if they came from a human, or trusted as such. That's some scary shit.

Show threadMartijn Faassen has movedDec 7, 2022

@pbrane @patrizia @andrewfeeney @kevlin @cwebber

It definitely gives the impression of an ego if you tell it that it's wrong. It tends to stick to its guns in a passive aggressive style: "sorry if I was wrong and new information may come to light but there's no evidence for X" when you just told it it's wrong about that.

Show threadJake MannixDec 7, 2022
@faassen @patrizia @andrewfeeney @kevlin @cwebber this is true. But *we* don't need to feel bad for telling it that it's wrong, at least.
Show threadKevlin HenneyDec 7, 2022
@faassen @pbrane @patrizia @andrewfeeney @[email protected] A mansplainer who's been on a soft skills course because he's been told quite firmly that he needs the training
Show threadEdwin LyonsJan 14, 2023
@kevlin @andrewfeeney @patrizia @pbrane @faassen @cwebber At least when I give it programming challenges, it is often quite responsive to my corrections. I use it for Unity (game) code sometimes and it has a tendency to make up functions that don’t exist. If you tell it that, it will handle that sensibly, though it can be a cat-and-mouse game if it then makes up another function that doesn’t exist!
Show threadKevlin HenneyJan 14, 2023
@edwin @andrewfeeney @patrizia @pbrane @faassen @[email protected] In other words, ChatGPT offers a way of automatically generating legacy code 🤔 Every piece of code generated by ChatGPT is, effectively, someone else's code, and needs to be understood anew.
Show threadEdwin LyonsJan 14, 2023
@cwebber @andrewfeeney @faassen @patrizia @kevlin @pbrane Yes, the way I see it is that it’s a bit like getting someone junior on my team to write code - I’m going to have to review it carefully and give feedback. But it is helpful with simple, well defined problems where the correct solution can be reached with a bit of back-and-forth.
Show threadMartijn Faassen has movedJan 14, 2023

@edwin @cwebber @andrewfeeney @patrizia @kevlin @pbrane

Github copilot functions as a kind of savant junior pair programmer who is very aware of patterns and idioms. It helps speed up dull stuff but also regularly introduces me to new idioms and APIs

Show threadMartijn Faassen has movedJan 14, 2023

@edwin @kevlin @andrewfeeney @patrizia @pbrane @cwebber

Interesting. I haven't used it for software; I use github copilot though which is less argumentative.

Show threaddyedgreyDec 8, 2022
@pbrane @patrizia @andrewfeeney @kevlin @cwebber
Do we really need more dis-/mis- or just generally bad information out here, just so some engineers can claim “10% more clever” for their 15 minutes? We should be using our power for good. Our ethics and mores are really not keeping up with the tech.
Show threadJake MannixDec 8, 2022

@dyedgrey @patrizia @andrewfeeney @kevlin @cwebber I... don't think this thing is there "just so some engineers can claim '10% more clever'". I think ChatGPT is an amazing breakthrough and is genuinely useful.

I also think it has to be treated really really carefully, as yes, it can produce vast quantities of fake news, bad answers, etc.

The invention of dynamite lead to lots of weapons of destruction. It is also genuinely useful, industrially.

But I agree: we should not forget the harms.

Show threadbigiainDec 10, 2022
@patrizia @andrewfeeney @pbrane @kevlin @cwebber On the other hand, I’ve worked with people who make less sense that ChatGPT… (I’m also vaguely considering an Xmas break project where I hook ChatGPT up to the work Slack channels, then get a second job while it impersonates me…)
Show threadPaul WeberJan 12, 2023

@patrizia @kevlin @andrewfeeney @cwebber @pbrane

Its *incorrect* outputs emitted onto an ignorant and unprepared world of users who might use unwisely use them.

I’m surely missing a key piece of information but, my own interactions with the machine and the results it has produced, as of today do not justify any kind of paid subscription (and apparently they’re planning a stock offer as well?…)

Show threadJames 🦉 #FBPE Jan 13, 2023
@pbrane @patrizia @andrewfeeney @kevlin @cwebber But we're not in control - a prompter might be at the moment of generation, but then once its output is used in articles, posts, papers and entertainment, the audience has very little to no control over it. It's similar to troll and bot farms in that respect. Not everyone has been directly influenced, meaning it's very hard for them to know they have been _indirectly_ influenced, ie through friends, family, and even the media, who were directly.
Show threadrootlessJan 14, 2023
@pbrane @patrizia @andrewfeeney @kevlin @cwebber plenty of humans that shouldn’t be emitting…
Show threadSylvDec 8, 2022
@patrizia For now, and given it seems people have been using it to flood stack overflow with wrong answers, that's not even sure @andrewfeeney @kevlin @cwebber
Show threadYaoi Box BalloonJan 10, 2023
@patrizia @andrewfeeney @kevlin @[email protected] I'm screaming.
Show threadJobJan 26, 2023
@patrizia @andrewfeeney @kevlin @cwebber Oh no… you just gave me a nightmare vision of an even more dystopian future where we have to endure "unsollicited automated mansplaining as a service"
Show threadKevlin HenneyJan 26, 2023
@vanderZwan @patrizia @andrewfeeney @[email protected] "The future is already here – it's just not evenly distributed." William Gibson
Show threadAndrew FeeneyJan 26, 2023

@kevlin @vanderZwan @patrizia @cwebber

#ClippyGPT

Show threadKevlin HenneyJan 27, 2023
@andrewfeeney @vanderZwan @patrizia @[email protected] 😄
Show threadAndrew FeeneyJan 27, 2023
@kevlin @vanderZwan @patrizia @cwebber I’m not going to lie, I had real trouble ordering the adjectives in this meme. Reading it again now it seems very wrong.
Show threadJobJan 27, 2023
@andrewfeeney @kevlin @patrizia @cwebber … have you tried asking ChatGPT?
Show threadAndrew FeeneyJan 27, 2023

@vanderZwan @kevlin @patrizia @cwebber

My first thought was: Determining the natural order of these given adjectives is actually the exact problem for which ingesting the entire internet into a generative model would be the insanely over-engineered but near perfect solution. Let's try it!

My second thought, after trying it, was that the results were ... inconclusive.

Show threadJobJan 27, 2023
@andrewfeeney @kevlin @patrizia @cwebber at least it keeps up the mansplaining tradition of confidently contradicting something it just said
Show threadAndrew FeeneyJan 27, 2023

@vanderZwan @kevlin @patrizia @cwebber

Okay fine, I'll give it this one. My prompt was wanting.

Show threadPatriziaJan 27, 2023

@andrewfeeney @vanderZwan @kevlin @[email protected] This video on the subject of adjectival ordering is worth watching :)

CppCon 2018: Borislav Stanimirov 'The Bad Big Wolf Meets Riding Hood Little Red' - https://www.youtube.com/watch?v=Dw0UBuTKHHg

CppCon 2018: Borislav Stanimirov “The Bad Big Wolf Meets Riding Hood Little Red”

http://CppCon.org—Presentation Slides, PDFs, Source Code and other presenter materials are available at: https://github.com/CppCon/CppCon2018—Lightning Talk—...

YouTube
Show threadAndrew FeeneyJan 27, 2023
@patrizia @vanderZwan @kevlin @cwebber I've bookmarked this for tomorrow morning. Looking forward to it! (Bed time now!)
Show threadJobJan 27, 2023
@patrizia @andrewfeeney @kevlin @cwebber The fact that you have a video ready for this topic just made my day
Show threadAndrew FeeneyJan 27, 2023
@patrizia @vanderZwan @kevlin @cwebber This video was brilliant!
Show threadgeobeckJul 28, 2023

@patrizia @andrewfeeney @kevlin @cwebber

Just don't check the "Actually..." option in Settings. 😆

Show threadJulian ElveDec 6, 2022

@andrewfeeney @kevlin @cwebber

@ton the essence of
@not explained ? 🙂

Show threadahistorical immaterialistDec 6, 2022
@andrewfeeney @kevlin @cwebber I have found, if you tell it that it's wrong, it will immediately, unreservedly correct itself, so it hasn't completely managed to emulate a man.
Show threadHans GerwitzDec 6, 2022
@andrewfeeney @kevlin @cwebber I’ve read many mansplaining comparisons and they’re accurate and funny. But until you put it this way hadn’t really seen the how instrumental the lack of perspective-taking is!
Show threadHans GerwitzDec 6, 2022
@andrewfeeney @cwebber @kevlin now I wonder if this is a good tool to teach men about what mansplaining feels like 😆
Show threadAndrew FeeneyDec 6, 2022
@hans @kevlin @cwebber Maybe, but whether it is or isn’t, I doubt you or I are qualified to say. 😅
Show threadHans GerwitzDec 6, 2022
@andrewfeeney @kevlin @cwebber I’m pretty sure I am very qualified to say what mansplaining feels like.

Oh, we were talking about the receiving side.
Show threadAndrew FeeneyDec 6, 2022
@hans @kevlin @cwebber Ha!
Show threadDavid EmmettDec 6, 2022
@andrewfeeney @kevlin @cwebber Automated Reply Guy
Show threadAndrew FeeneyDec 6, 2022
@motomatters @kevlin @cwebber I fed this thread into ChatGPT and this is what it said:
Show threadAndrew FeeneyDec 6, 2022
"The term "automated reply guy" could refer to a number of different things, but it is typically used to describe a person or system that automatically responds to messages or queries without truly understanding the content or context of the conversation. 1/2
Show threadAndrew FeeneyDec 6, 2022
This could be seen as a form of automation that lacks the ability to exhibit true intelligence, as it does not have the ability to understand or engage in a meaningful way. Instead, it relies on pre-defined responses that may not always be appropriate or relevant to the conversation at hand." 2/2
Show threadKevlin HenneyDec 6, 2022
@andrewfeeney 😄
Show threadFrancesco BiscaniDec 6, 2022

@andrewfeeney @kevlin @[email protected]

> Mansplaining As A Service

No wonder ChatGPT is so popular on HackerNews!

Show threadMike Taylor 🦕Dec 6, 2022
@andrewfeeney LOLOLOL
Show threadCharlie StrossDec 6, 2022
@cwebber @kevlin @andrewfeeney For my new startup I am going to license the rights to the "Boris Johnson" brand name and market my rip-off of ChatGPT as "Bullshit Boris™" (or maybe "🇬🇧🇬🇧🇬🇧 Brexit Bloviator 🇬🇧🇬🇧🇬🇧" if the UK market will support it).
Show threadKevlin HenneyDec 6, 2022
@cstross @[email protected] @andrewfeeney 😄
Show threadKroshnit🇺🇦Dec 6, 2022

@andrewfeeney @kevlin @cwebber

The original name was in fact "Ack-ChatGPT-yually"

Show threadTim Trautmann 📷Dec 6, 2022
@andrewfeeney @kevlin @cwebber chef's kiss.
Show threadElWollito Dec 6, 2022
@andrewfeeney @kevlin @cwebber Alway add a giraffe to the question and it will readily implement it in its answer