“Mansplaining as a service.” Spot on.

Quoting @andrewfeeney: https://phpc.social/@andrewfeeney/109466122845775778

Andrew Feeney (@[email protected])

@[email protected] Christine Lemmer-Webber (@[email protected]) 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.

PHP Community on Mastodon
Richard Gaywood (@[email protected])

Attached: 1 image me, rekt, standing in the smoking ruins of the point i was trying to make as i slowly shrink and transform into a corncob: “deep lesson in this”

Infosec Exchange
@stevenodb 🤣🤣 geweldig dit
@stevenodb Tell me you're insecure without telling me you're insecure.
@Roelofcliffortvos @stevenodb I didn’t do anything. Why are you always picking on me? I’m going to talk to the manager about this, to recommend that a committee investigate why you’re asking questions.😂
Now at [email protected] ! (@[email protected])

oh shit it's mansplaining as a service

the Octodon
@terrie @stevenodb I literally posted a link to this toot the exact same minute you did! 😅 Should have done it much earlier of course!
@stevenodb @andrewfeeney even though I don't quite agree, this did give me a chuckle. Certainly feels that way at times!
@stevenodb @andrewfeeney Spouse (who is an expert on certain topics): George Santos and I get the same weighting in their training set (never mind that they're supposed to pay me).
@andrewfeeney @stevenodb Good call. I’ve also found that it makes up facts & bluffs them confidently.
@stevenodb @andrewfeeney my hot take is that I found ChatGPT very helpful in organizing & outlining my ideas. I have weak organizational skills & want to write a memoir. The amount of data was staggering. AI helped to herd my psychological cats.

@stevenodb @PavelASamsonov @andrewfeeney

Unless ChatGPT can cite every single sources that it got its information on, it's pretty much useless.

However, that has already been the problem of writings (especially Medium) on the Internet for the past two decades. Very few writers care about citing research and evidences to back up their points anymore.

@stevenodb @andrewfeeney VC firms moaning with pleasure at the thought of getting in on the ground floor of investing in “MaaS” 😂
@stevenodb @andrewfeeney
It’s funny because it’s true 🤣
@stevenodb @andrewfeeney This is too accurate! I asked it to generate an Ansible playbook using a module I’m quite familiar with. Not only was the playbook wrong, it was also providing parameters that do not exist in the module! 🤦‍♂️
@stevenodb @andrewfeeney Sounds vaguely Libertarian as well.
@stevenodb @andrewfeeney with insight always taken from someone more qualified.

@stevenodb @andrewfeeney i used to tell people that if you say random stuff without regards to its truth with a straight enough face that people will believe you. I just call it messing with people.

for instance, i had people going for years with the thought "have you ever laid in bed, trying to sleep, and your body just jerks? That's your body dumping useless information- that's why you can't remember some stuff the next day"

AND THEY BOUGHT IT

@stevenodb @andrewfeeney I was thinking along the lines of Indian WhatsApp Uncle but this is also good
@stevenodb @andrewfeeney No better example than asking it to make an amigurumi crochet pattern. People have made the patterns. Completely bizarre.
@stevenodb @andrewfeeney it’s like a robot Internet forum except that there’s no one to correct the wrong answer

@stevenodb @andrewfeeney

Just like in human communication one is responsible to let the other party know ones level of expertise.
If one appends "Explain like i am five.", "Explain like i took a course on that in university but i need an refresher" or something similar one can tune the kind of response.

Prompt tuning is a thing and it is frankly annoying that the same crowd both dismisses results achieved by prompt tuning, that prompt tuning is a skill and complain

@stevenodb @andrewfeeney about defects in outputs which could be eliminated by some prompt tuning.
Yes by prompt tuning one injects oneself into the process. This feedback loop is nothing novel https://en.wikipedia.org/wiki/Evolutionary_art
Evolutionary art - Wikipedia

@stevenodb @andrewfeeney But then again non of the AI maximalists describe it as:
You can spend 2 hours tweaking the resulting images, seeds and the prompt to democratize the commissioning of images or have 10 different conversations about the same thing where you strategically walk down different paths of conversation to get a better approximation to debias the conversation.
... Which would be the honest description
@stevenodb @andrewfeeney if not done by a man, is it just splashing? You wouldn't call a woman who explains without ascertaining a man splainer. Likewise chatgpt is also not a man, just a splainer. The spot onness of the description doesn't make the description useful though.
@stevenodb @andrewfeeney I’m stealing your post, but telling you in advance because of manners.😂

@stevenodb @andrewfeeney I just wrote about ChatGPS quoting the line that it is the world's ultimate bullshit artist.

"Mansplaining as a service" is even better.

@stevenodb @andrewfeeney AND every single piece of data it's throwing at you is stolen from someone else's hard work.
@stevenodb @andrewfeeney It's been entirely unsurprising watching how many shit-tier guys this has pissed off. The mere mention of mansplaining as a negative thing brings the mansplainers out of the woodwork and this toot from Andrew Feeney got them so outraged and... unsurprisingly mansplaining
@stevenodb @andrewfeeney
I was reading a Reddit thread about someone getting a fail for using ChatGPT to write a paper.
@stevenodb @andrewfeeney has chatgpt never been heard to say " I don't know"?
@msokolov @stevenodb This will be the new captcha test. To pass the test you must answer “I don’t know” to the prompt question otherwise we assume you are a generative model.

@andrewfeeney @msokolov 😁 — it is known to deny specific topics and it won’t for instance make up a narrative about how iOS is better than Android. But there’s workarounds for both.

Never seen “I don’t know”. I know you’re kidding, but ChatGPT’s goal is not to answer questions, but to engage in human-like dialog.

@stevenodb @andrewfeeney May our Future President and Glorious Leader ChatGPT have mercy on us all.
@jaramsli @stevenodb I for one also welcome our new robot overlords.