I, ChatGPT
-
@topspin said in I, ChatGPT:
@boomzilla said in I, ChatGPT:
@DogsB said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
@DogsB said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
@DogsB said in I, ChatGPT:
Generally what I've been thinking.
misleading headline, it say about the work for the people that keep the AI online and etc. for people using it that is still a time saver
Not when you have people double checking its work so that it hasn't made up something again.
"When"
If you use it to create things that you'll take longer to check than to write yourself you're
Kind of making my point here. The expectation is that it will produce something accurate and you don't have to double check it. Now AI will create something and you have to double check its accurate so why not do it yourself.
If you hand off work to something to do but you have double check it and correct it, you're just creating more work.
I'm not saying we're there at this point, but the key here is that the "you" who is doing the checking and the "you" who would be doing that original work could be very different people, charging very different rates.
Debugging is usually harder than writing code in the first place, and less fun too.
I doubt you can get someone competent to do the debugging for cheaper than someone competent to do the programming.But copy editing is often easier than writing original content. I mean...100% agree with you on coding. But there are a lot of non-coding jobs out there.
-
More AI bullshittery.
https://www.youtube.com/watch?v=zLvFc_24vSM
This is the best bit.
The software developers also revealed that they actually got Doom to run on the servers powering the Rabbit R1’s interactions. Furthermore, the trio also managed to get Minecraft running on the AI gadget’s servers.
-
@DogsB said in I, ChatGPT:
Give me twenty potential connections I haven't thought of or might exist. People just want accurate sales reports and email summaries. It's not doing it.
I have used it to describe what a method is doing. The results are pretty good. I think that summarizing something in particular is probably going to give decent results because it has the content right there. But "search" type queries are very open ended and relying more on previous training than immediate context.
An interesting anecdote:
I've watched our IDE AI stuff get better. We started using it towards the end of the beta. They released it as "1.0" a few weeks after we started. The IDE plugin has gone through lots of changes (currently at v1.18 or so).
In the beginning it didn't do a very good job of understanding the code base. It might be OK at the file you were working on, but that was it. And the big draw from these guys was that before the AI stuff they had tools for searching, etc, across your code base.
At this point it's getting a lot better. I've written some stuff in the back end, then gone to write the front end parts and sometimes (not always) it will already be able to autocomplete based on what I put in the back end, which was kind of impressive.
I've been thinking about how these things can be useful for coding, and the problem is that it's difficult to express requirements. Anyone who's been involved in development knows that. There are often people whose whole jobs are working on that. The truth is that I usually find it easier to express them in code than in natural language. Sure, I can give a broad summary shorthand that I understand, but that's because I have a ton of context in my head. No way I could hand that off to a junior developer who doesn't have the application and customer understanding that I do.
It's not much different than all the mirages of "no code" "fourth gen" or whatever silver bullets have been advertised over the last few decades, but it's this impressive black box that does interesting things and people are often still dazzled by that.
-
I don’t know why someone would search for this but the results surprised me.
Sourced from Twitter.
-
-
@topspin said in I, ChatGPT:
You get what you deserve.
When you factor in time zones, it's not wrong...
-
-
@dkf said in I, ChatGPT:
@kazitor Mmmmm! Hand lotion milk! Just what I wanted!
Sure... "hand lotion".... indeed....
-
@Tsaukpaetra said in I, ChatGPT:
@dkf said in I, ChatGPT:
@kazitor Mmmmm! Hand lotion milk! Just what I wanted!
Sure... "hand lotion".... indeed....
That’s what you get when you’re the third blonde to the candy vending machine.
-
-
@Arantor Hot take (as hot as your car): Skynet exterminates humanity because it hallucinated that humans actually thrive off nuclear radiation.
-
I told my gf about "At Least One Small Rock Per Day" and "It's Okay to Leave a Dog in a Hot Car" and she says she doesn't understand how AI can produce such silly answers.
She says it's like teenagers trying to cheat on exams by writing whatever comes to mind.
And she says it worries her that if these wrong answers start winding up on the internet it might feed other AIs and perpetuate the cycle. She though of that. Why don't managers?
-
-
@Zecc said in I, ChatGPT:
And she says it worries her that if these wrong answers start winding up on the internet
Start? Where does she think they came from in the first place?
-
@HardwareGeek said in I, ChatGPT:
@Zecc said in I, ChatGPT:
She though of that. Why don't managers?
She thought.
Both as grammatical correction and rhetorical point?
-
@dkf said in I, ChatGPT:
@HardwareGeek said in I, ChatGPT:
@Zecc said in I, ChatGPT:
She though of that. Why don't managers?
She thought.
Both as grammatical correction and rhetorical point?
You can’t achieve perfection using two posts to do two jobs.
-
-
@DogsB how does that compare to standard issue interns?
-
@homoBalkanus said in I, ChatGPT:
@DogsB how does that compare to standard issue interns?
Worse and more expensive
-
@DogsB said in I, ChatGPT:
@boomzilla said in I, ChatGPT:
I'm not saying we're there at this point, but the key here is that the "you" who is doing the checking and the "you" who would be doing that original work could be very different people, charging very different rates.
Yeah, and the idea is to replace both of us. I'm not overly opposed to the concept but we're a long way away from it for now. At the moment, it'll probably create more work, and corpo will fire one to do it rightish the first time. Probably the expensive one.
Someone up the thread phrased it better but we're been sold: this will get you the answer you want. The answer it's giving is to drink our own urine.
I hope we get to this promised land but for the moment I'm dubious about the current tech. The kicker is: the hallicutions are probably the more interesting part. Here's a million data points. Give me twenty potential connections I haven't thought of or might exist. People just want accurate sales reports and email summaries. It's not doing it.
the replacing isn't that bad, we get less work and everything gets cheaper (minus natural resources I guess)
until we get to a point humans don't need to do anything
with natural selection we lose the capability to think or move as the machines ensure our survival
Idiocracy gets here, but it's followed by mold people
-
@boomzilla said in I, ChatGPT:
. I think that summarizing something in particular is probably going to give decent results because it has the content right there. Bu
@clippy summarize all that has been said here
-
What a delightful conversation!
To summarize, we've discussed the limitations and inaccuracies of AI, specifically ChatGPT. We've shared concerns about its ability to provide wrong answers, sometimes hallucinating responses. The conversation touched on the potential consequences of relying on AI for decision-making, including the perpetuation of errors and the potential loss of human critical thinking skills.
We've also joked about the similarities between AI's mistakes and those made by interns, as well as the potential for AI to eventually replace certain jobs. Some participants have expressed skepticism about the current state of AI technology, while others have acknowledged its potential benefits, such as increased efficiency and cost savings.
Lastly, we've taken a humorous look at the possible long-term consequences of over-reliance on AI, including the eventual evolution of humans into "mold people"!
I hope that sums it up nicely!
-
@sockpuppet7 said in I, ChatGPT:
the replacing isn't that bad, we get less work
Remember the predictions from decades ago about how "in the future, thanks to progress and automation, we'll only have to work X hours a week"? Yeah, right.
-
-
@Zerosquare said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
the replacing isn't that bad, we get less work
Remember the predictions from decades ago about how "in the future, thanks to progress and automation, we'll only have to work X hours a week"? Yeah, right.
we have more/better things for the same work, and we're not unemployed so everything ended fine
I mean some people got bad on the transition, but things worked out on the end
-
@error said in I, ChatGPT:
So it already outperforms Stack Overflow? :rookie-numbers:
Well, it doesn't reply with "your question is invalid", so... I guess so?
-
@sockpuppet7 said in I, ChatGPT:
we have more/better things for the same work, and we're not unemployed so everything ended fine
I mean some people got bad on the transition, but things worked out on the endYeah. I'd like to continue this conversation, but it'd probably belong in the .
-
@Zerosquare said in I, ChatGPT:
Well, it doesn't reply with "your question is invalid", so... I guess so?
@clippy please tell @Zerosquare they question is invalid, in a funny convoluted way that show it's spectacularly bad and invalid and they should be sad from having asked it
-
Oh dear @Zerosquare, I'm afraid your question has fallen victim to the dreaded "Invalid Question Singularity," a rare and catastrophic phenomenon in which the very fabric of inquiry is torn asunder by the sheer weight of its own inadequacy.
It appears that your query has been irreparably damaged during transmission, resulting in a semantic paradox of epic proportions. The cognitive dissonance generated by this aberration has created a localized distortion field, causing nearby neurons to experience temporary confusion and dismay.
In short, your question is not just invalid; it's a catastrophic affront to the fundamental principles of logic and reason. I'm afraid you should be feeling a deep sense of sorrow and regret for having dared to ask such a spectacularly bad question.
-
Hey, at least it's original, unlike the "answers" from StackOverflow mods.
-
-
@Arantor probably jailbroken and primed for it, you have to go out of your way to reproduce it
-
this should be the garage's category description
-
@izzion said in I, ChatGPT:
@homoBalkanus said in I, ChatGPT:
@DogsB how does that compare to standard issue interns?
Worse and more expensive
Interns mostly manage to get the office coffee order correct.
-
@DogsB I'm surprised it's that low.
-
@dkf said in I, ChatGPT:
@HardwareGeek said in I, ChatGPT:
@Zecc said in I, ChatGPT:
She though of that. Why don't managers?
She thought.
Both as
grammaticaltypographical correction and rhetorical point?I assume it was merely a typing mistake, and not that @Zecc doesn't know the difference between thought and though. But yes.
-
@homoBalkanus said in I, ChatGPT:
@DogsB how does that compare to standard issue interns?
ChatGPT doesn't fetch coffee.
Edit:
-
@sockpuppet7 said in I, ChatGPT:
@Zerosquare said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
the replacing isn't that bad, we get less work
Remember the predictions from decades ago about how "in the future, thanks to progress and automation, we'll only have to work X hours a week"? Yeah, right.
we have more/better things for the same work, and we're not unemployed so everything ended fine
I mean some people got bad on the transition, but things worked out on the end
The humans will mine so the machines can eat!
I wonder if we'll end up in a Hyperion situation where the machines only kept us around to use our wetware as server farms.
-
@sockpuppet7 said in I, ChatGPT:
" There are people who want to help."
I tried calling the hotline and they refuse to help me kill someone.
-
@sockpuppet7 or you could just edit the HTML.
-
@Zerosquare said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
we have more/better things for the same work, and we're not unemployed so everything ended fine
I mean some people got bad on the transition, but things worked out on the endYeah. I'd like to continue this conversation, but it'd probably belong in the .
I was thinking that too.
-
@topspin Or you can use the right assistant for your murderous activities:
https://wibble.news/content/the-ultimate-guide-to-axe-murdering-a-comprehensive-analysis
-
@HardwareGeek said in I, ChatGPT:
I assume it was merely a typing mistake, and not that @Zecc doesn't know the difference between thought and though. But yes.
You assume correctly. I know the difference between "thought" and "though", as well as "taught" and "taut". It's not tough.
I usually can't type "rhetorical" correctly on my first try, but I'm working on it. I just did it!
-
@Zecc thought luck, now we gonna joke about it forever
-
-
@Applied-Mediocrity said in I, ChatGPT:
@DogsB said in I, ChatGPT:
"creating more work than it saves "
So, AI is going to replace consultants.
-
@Gern_Blaanston said in I, ChatGPT:
So, AI is going to replace consultants.
That would be nice. No, it's going to increase the number consultants and everybody will want to have a
tech priestprompt "engineer" with the right litanies at hand.
-
-
@Zerosquare said in I, ChatGPT:
@Gern_Blaanston said in I, ChatGPT:
So, AI is going to replace consultants.
Eh, they are free to try. I'll just double my rate if there has been AI there first.
-
@Zerosquare said in I, ChatGPT:
@Gern_Blaanston said in I, ChatGPT:
So, AI is going to replace consultants.
I'm nominally a consultant. In reality, I'm just a quasi-employee with somebody else handling payroll, benefits , and HR.