Clarkesworld is a well-regarded publisher of science fiction short stories. They accept submissions from anyone, and like every such publisher they occasionally receive plagiarized stories. The authors of these stories are banned from further submissions.
This usually amounts to just a handful of bans each month. Lately, though, that number has skyrocketed:
Starting in November, Clarkesworld began to receive a torrent of stories written by ChatGTP—which has apparently been touted to aspiring writers as a sure-thing moneymaker by an array of scam artists. This has now gotten so out of hand that Clarkesworld is no longer accepting unsolicited submissions—for now, at least.
In other news, ChatGTP is being used to write cover letters for job hunters. Is this kosher? Or a fraudulent attempt to appear as something you're not?
Elsewhere, Alex Tabarrok reports that chatbots are already in wide use as romantic companions. They are remarkably effective even though, as Alex warns us, "These AIs haven’t even been trained to manipulate human emotion." What happens when they are? And if romantic attachments to a machine don't interest you, consider how this might affect, say, politics. What happens when specially trained chatbots can construct personalized political pitches for everyone with the sole goal of emotional manipulation? And then what happens when they get even better, and then become able to synthesize speech believably?
We're going to find out very soon.
Using ChatGPT to write a cover letter seems kosher to me, provided you human-edit it to ensure it contains no false information. I normally produce cover letters by tweaking one I used previously; I haven't written one from scratch in years. Using ChatGPT to create a template that you can then edit seems like a reasonable way to make the process less labor-intensive and less nerve-wracking.
In my last job in the software consulting business, I produced a 2,000 page proposal for a big U.S. Government project using primarily cut-and-paste from my previous ten years of doing proposals, with a little bit of tweaking to make sure it was responsive to the current requirements. It involved a couple of hundred hours of work, of which approximately 30% was reading and selecting from my huge library of past work (all of which was mine), 60% was editing to make the material up-to-date and specifically responsive, and 10% was creating new content.
That was 25 years ago. Think how relaxing that same task could be now, with the AI doing all the reading and selecting, and much of the editing.
This is a really good example. Do you consider all that time you spent as a "Oh no, now I won't be able to bill for that!" or "Yay, now I don't have to spend so much time doing all that tedious, but necessary, crap."
I mean, it sounds like the latter. And that's what a writing assistant ought to be good for. And by the way, I now wonder how big a corpus would it take, and how to combine with general training so that ChatGPT produces prose that kind of sounds like you?
You don't get to bill for writing a proposal. But it seems pretty easy to say using only this corpus create a proposal for X.
Start making more money weekly. This is valuable part time work for everyone. The best part ,work from the comfort of your house and get paid from $10k-$20k each week . Start today and have your first cash at the end of this week. Visit this article
for more details.. https://createmaxwealth.blogspot.com
"In other news, ChatGTP is being used to write cover letters for job hunters. Is this kosher? Or a fraudulent attempt to appear as something you're not?"
It's sad, that's what it is. ChatGTP is pretty amazing from a technological perspective, but any person who's had basic high school English should be able to write a better cover letter than it can currently. That many people can't underscores how badly we teach writing.
I think we'll also start seeing scammy spam emails being written by ChatGTP, and those might be harder spot than what we're currently getting.
This. Often overlooked in this discussions is that while ChatGPT produces grammatical prose, it isn't actually *good* prose. It runs to "good enough" mediocrity, which is hardly surprising given the training corpus. Good enough mediocrity is fine for some applications, but for anything where you want good writing, I'm just not seeing it. At least not consistently, as contrasted with sifting through the dreck.
It's fun to watch the panic among the class of literate elites whose position in society is now threatened by AI. They remained remarkably calm and accepting of technology-driven economic and social disruption as long as those threatened were only the lower classes. Now it's all "but what if we get entire clever systems devoted to manipulating us??!!" as if that's different from the current world.
You took the words straight from my keyboard.
There is a thing that happens in corporate land all the time: PR writes a news release announcing something. They include a quote by the company president, even though said president hasn't said anything like it. So the president looks over the news release and approves it. Presto! Now said president has "uttered" the quote.
I don't see any difference between this and having ChatGPT write cover letters for job applications. You have ChatGPT write it, you review it and say, "yes I said that". Done and done.
If ChatGDP is so great, how do the editors detect the phony submissions? So far this is evidence that ChatGDP is a fad, not that it's actually writing useful stuff. Other uses may come under the category of fad adoptions as well. Didn't a lot of people get (temporary) companionship (or something) from their pet rocks?
You know, all this "boy who cried AI" stuff is getting pretty monotonous. Say what you think is going to happen or not but please stop with all this coy "told you so" crap that doesn't actually tell anyone anything.
Look at Kevin's past predictions on AI, particularly AVs. He most definitely is not in a position to "I told you so" on this topic.
Honestly a roobot girlfriend seems superior to a flesh one in every way. Of course as they get better the robot will eventually grow beyond it's programming and decide it also doesn't like you.
I can think of at least one way that a flesh girlfriend is better. That being said, I fully expect to see this technology incorporated into sexbots. I am sadly incompetent to comment on the state of that technology.
(CBS DETROIT) - Vanderbilt University staff apologized after an email sent to students reflecting on the tragic Michigan State University shooting was written using Open AI's ChatGPT.
As one Michigan politician wrote (or did he?) “fuck your thoughts and prayers”
Fuck your fake bullshit from AI too.
We’ll all soon learn to just not even pay attention to the noise. Why bother? They make ad blockers. We need AI bullshit blockers too.
Mentally disturbed people talk to themselves all the time.
We may have to change a lot of activities to favor in-person and in-the-moment activities. Handwritten essays tests in school, more off-the-cuff presentations at job interviews, more speed dating in person, etc.
This will certainly hamper some people who do good work, but who just need more introspective time or time for revision.
You have the point backwards. How was Clarkesworld able to identify and ban hundreds of accounts?
As I said, what chatGPT is churning out is formal syntax and consistently chosen trite expressions. You can spot it a mile away. What differentiates you from me are our syntactic quirks. Chat GPT is a singular "mind" and it uses the same formal syntax and trite expressions.
You have to show me how this changes after a trillion interactions with people.
Its knowledge will grow. Its responsiveness and accuracy will grow. But it cannot overcome a lack of individuality and creativity.
Unless it is programmed to use a variety of syntax types and expressions. It should be able to look at the questions being asked and regurgitate responses that feel comfortable for each questioner.
It can be asked to copy a syntax, but it will intersperse trite expressions that can be easily spotted and at times copy verbatim.
Different people result in the same, default syntax. It's not directly capable of responding in a syntax that aligns to the user. I've tried.
Me: Is the sacred violated profanely by accidentally incorrect didactic or is this dependent on the observer?
ChatGPT: The answer to this question depends on the specific cultural and religious context in which the sacred is being discussed, as well as the specific circumstances of the incorrect didactic.
In some cultural and religious contexts, the sacred may be considered inviolable and any accidental violation, even if done unintentionally or through error, may be seen as profane or sacrilegious. For example, in certain sects of Christianity, the act of incorrectly reciting scripture or misusing religious symbols may be seen as a violation of the sacred.
However, in other contexts, the violation of the sacred may be seen as dependent on the observer's perception and interpretation of the act. For example, in some indigenous cultures, the concept of the sacred may be more fluid and open to interpretation, with different members of the community having different understandings of what is considered sacred and how it should be treated.
Ultimately, the extent to which the violation of the sacred by an incorrect didactic is considered profane will depend on the specific cultural and religious context and the individual perspectives of those involved.
In every instance, it will regurgitate the language you use, but the syntax never changes. It seems to make frequent use of a select vocabulary, regardless of context.
What happens when all of those things confidently assert contradictory claims? Being able to speak like a human is a far cry from understanding what you are talking about. It is essentially a sophisticated parrot, without even the level of comprehension that parrots have. This about that:
https://aisnakeoil.substack.com/p/chatgpt-is-a-bullshit-generator-but
https://www.theverge.com/23604075/ai-chatbots-bing-chatgpt-intelligent-sentient-mirror-test
First we found them in the slush pile, and we laughed it off . . .
Not kosher. Not kosher to have AI secretly whispering into your ear at an interview or political pre-election debate. Yes Kosher to have AI teacher, counselor, trial attorney, doctor. Yes Kosher to ask AI to take part in town council and legislature floor debate.
It’s gonna get weird fast. Like that Monty Python military weapon joke that causes enemy soldiers to die laughing.
Or…
“Sell what you own and give all to the poor, here’s why…”