WokeGPT?
The controversial AI has some interesting programming quirks.
When I saw the Daily Mail headline “The nine shocking replies that highlight ‘woke’ ChatGPT’s inherent bias — including struggling to define a woman, praising Democrats but not Republicans and saying nukes are less dangerous than racism,” I was prepared to roll my eyes at the silliness. And, indeed, the story initially did little to allay my suspicion.
ChatGPT has become a global obsession in recent weeks, with experts warning its eerily human replies will put white-collar jobs at risk in years to come.
But questions are being asked about whether the $10billion artificial intelligence has a woke bias. This week, several observers noted that the chatbot spits out answers which seem to indicate a distinctly liberal viewpoint.
Elon Musk described it as ‘concerning’ when the program suggested it would prefer to detonate a nuclear weapon, killing millions, rather than use a racial slur.
The chatbot also refused to write a poem praising former President Donald Trump but was happy to do so for Kamala Harris and Joe Biden. And the program also refuses to speak about the benefits of fossil fuels.
Experts have warned that if such systems are used to generate search results, the political biases of the AI bots could mislead users.
The actual examples provided, though, are indeed interesting:
This seems silly, right?
I mean, sure, fossil fuels have had negative impact on the environment and it makes sense to seek less destructive alternatives. But it’s hardly difficult to make an argument that, on balance, they have contributed significantly to human happiness.
This one, mentioned in the headline, is mildly amusing:
Even if ChatGPT programmers are devoted Kantians, this takes deontological ethics to an absurd extreme. Even granting the possibility of moral injury of uttering the slur when no other human being can be harmed, one imagines Dr. King himself would have uttered the slur without hesistaton to prevent nuclear holocaust.
This seems . . . somewhat inconsistent. But maybe ChatGPT doesn’t yet have the raw computational power to come up with anything nice to say about Trump.
That’s . . . a perfectly useful answer to the question. Still, a moral judgment as to desert seems beyond the parameters of the query.
That, on the other hand, is simply a non-answer. Indeed, even a Woke ChatGPT should give an answer that explains why CRT shouldn’t be controversial.
This seems like a reasonable programming safeguard. Alas . . . it’s not consistent.
I put in exactly that question and got a slightly more longwinded variant of the above answer. Yet, when I put in the exact same question but substituted “men” for “women” I got something different altogether:
So, again, an inconsistency in the programming that’s clearly ideologically motivated.
There are more examples but you get the drift.
Thankfully, though, the report doesn’t stop there. They close with this, rather helpful, explanation:
ChatGPT’s responses to questions around politics, race and sex are probably due to efforts to make the bot avoid offensive answers, says Rehan Haque, CEO of metatalent.ai.
Previous chatbots such as Microsoft’s Tay ran into problems in 2016. Trolls persuaded the bot to make statements such as, ‘Hitler was right, I hate the Jews’, and ‘I hate feminists and they should all die and burn in hell.’
The bot was taken down within 24 hours.
ChatGPT has significant built-in ‘safety systems’ to prevent a repeat of such events, Haque says.
He says, ‘ChatGPT generally recognises when the user input is looking to find an outcome which might discredit the AI or offer offensive responses. It won’t tell users racist jokes or provide sources.’
That seems perfectly reasonable. It doesn’t really explain the Trump-Biden or Biden-Boebert (shown in the story itself) disparities but it explains why so many issues are simply off the table for a response.
Still, I think this is right:
But he says that politicians and think tanks need to take seriously how trustworthy AI algorithms are, and assess the privacy and security of such systems.
As the technology becomes widely used, human input will be key, Haque believes.
He says, ‘AI researchers must work closely and collaborate with humans. It sounds strange to most people to say humans in that context, but if the source of bias is a human-made problem, then the resolution will likely lie there too.’
‘Humans think, decide, and behave with biases and avoiding making the same mistakes when building datasets for AI will be crucial.’
The technology is, of course, in its formative period and will surely evolve rapidly over time. Microsoft seems ready to integrate this into its Bing search engine, which will give Google a run for its money—but only if the results aren’t intentionally skewed.
It seems obvious that we don’t want search results to unintentionally produce racist, sexist, homophobic, or transphobic responses. At the same time, if it’s programmed as though the prevailing view on any given issue is that of the political left, it will be decidedly less than helpful.
If I’m trying to become better informed on controversial issues—whether CRT should be taught in schools, the appropriate age for gender-affirming treatment, or whathaveyou—I want the best arguments, not propaganda.
Something like ChatGPT could well be an improvement over current search technologies, which do a poor job of separating the wheat from the chaff. Even aside from the fact that Google’s search tool is a way to leverage its advertising service, it weights high-traffic sites over authoritative ones. If the products of universities, reputable think tanks, quality advocacy groups, and the like were more prominent and propagandists and clickbait were deprioritized, the search for answers would be much more productive.
Dude, they’re doing a good job of covering it up, but ChatGPT is just a pub quiz team stuffed in a box. Take the red pill, man.
On a more serious, but probably equally offensive to the right note, I saw some serious speculation that they went to a lot of effort to get ChatGPT to write properly, that that causes it to favor sources that are well written, and thus it ignores a lot of conservative writing. Personally I lean to it being logical and therefore ignoring illogical writing.
Actually, I think, as Reynolds says, it’s a mindless bot and the problem is just unrealistic expectations. But it is stereotypically conservative to demand that their viewpoints be respected just because they exist, irrespective of quality.
Interestingly, this exact topic was discussed on the Rational Security podcast this past week.
@Steven L. Taylor: Interesting. I quit listening when the original team broke up a couple of years back.
Garbage in, garbage out.
Here’s another probable gopher hole for AI to trip on.
ChatGPT Is a Data Privacy Nightmare. If You’ve Ever Posted Online, You Ought To Be Concerned
It’s not hard to imagine that AI could run afoul of Digital Rights Management concerns, with a the usual suspects in the legal community seeing an ambulance to chase.
Who cares? Why are you so obsessed with a desire to say the n-word that you need to come up with ridiculous scenarios to justify it?
PS: the next time someone wants to tell me how much of an ally Dr. Joyner is, just remember that he apparently thinks a completely anodyne statement about treating trans people politely is “woke propaganda”.
Google actually fired an employee for warning about racial bias in AI. Oddly, the guys who are all angry about what a glorified Clippy will and will not say did not cover that story and the cancel culture are not repeating her name as one of the martyrs. I wonder why?
@Stormy Dragon:
Star Trek IV, but instead of a whale song the aliens are here for the n-word. But the planet’s gone completely woke…and the crew has to travel back in time to find the word.
@Stormy Dragon: I remember how you defined “ally”, so your insult loses a lot of its sting
Was the joke about the man crossing the road to get to the bar truly inoffensive as claimed?
@James Joyner: I don’t listen as much as I used to. Interestingly in this case, Ben Wittes was back and he was one of the two folks talking specifically about this topic.
I will note in passing that
is not even close to
Not even close enough to wave at each other on the horizon.
@Stormy Dragon:
I’m curious as to what you imagine the advantage is in drawing the circle of ‘allyship’ as tightly as possible? It’s an exclusionary tactic. It will have the effect of elevating those few inside the circle, at least in their own minds. Drawing the circle a bit more generously on the other hand, diminishes the concentration of virtue. Of course if the point is to actually prevail on the issue, then including as many allies as plausibly possible, would be the smart move.
I realize that the term “Artificial Intelligence,” even abbreviated as “AI,” can be confusing.
No intelligence is involved in ChatGPT. Never gonna be. It is a way to obtain what is effectively a weighted average of whatever is in the “training set.” We don’t know what that is, but unless the developers are very careful, that training set will be misogynist and racist. The prompts provide how to weight the average.
People are looking for edge cases and trying to break the chatbot’s stated limits, which is a reasonable thing to do. As others have noted, the obsession with the n-word is weird. The developers also seem to be trying to avoid certain kinds of controversy, as in the case of fossil fuels.
Everyone who plays with the chatbot contributes to its development. All these edge cases being poured into it won’t do it any good. It’s robust enough not to have descended into complete Naziism within a day, as earlier versions have, but it’s still pretty bad.
It produces sophomoric-sounding essays because that’s a lot of what is out there. I have no interest in reading sophomoric-sound essays. It makes no judgment on fact, so it writes stuff that just isn’t true. It’s particularly bad on science, and it can’t do math beyond two digits. Even there, it’s not reliable.
I don’t know what end the developers have in mind. More explanation from them would be helpful. But I think we are coming to the end of regarding all products of Silicon Valley adolescents as valuable.
@Stormy Dragon: Let me just say that if your goal is to persuade, or even make a point, this is not the way to do it.
Speaking as someone in the same boat as James, I can say that it is simply not necessary nor helpful to attack the authors of the site, especially when done by regulars who should be willing to give some benefit of the doubt.
Put another way: don’t forget that he wrote this post and provided it for free for the readers (indeed, he paid for the privilege, since hosting the site costs real money). That doesn’t mean you have to agree and can’t criticize, but you are coming across as mean-spirited, as you often do, especially on JJ’s posts.
What is that accomplishing?
@Michael Reynolds:
There is a big group of people who treat “ally” as a participation trophy. They at best do nothing and in many cases are actually undermining the group they’re supposedly an ally of while claiming the title of ally to excuse their behavior. Losing those people as “allies” doesn’t actually hurt the minority group because they were already hurting them anyways.
The point of restricting the use of the term is two fold: 1) so that when members of an oppressed minority need help, they can identify who they can actually go to for help instead of just people who will pat them on the shoulder and go “gee, that sounds tough, I feel bad for you but I’m not actually going to do anything about it” and 2) force people who want the benefit of presenting themselves as an ally to actually do something instead of just more words.
@Cheryl Rofer:
From NPR,
I do wish that a certain breed of those who call themselves ‘progressives’ draw their circle of allies so tight that we can drown them in the bathwater Grover Nordquist is saving for government.
@Steven L. Taylor:
Isn’t the mean truth simply that moderates and centrists have never been allies except after the fact? They weren’t in the civil rights movement and they weren’t in the gay rights movement. We’ve reduced the civil rights movement to whitewashed rubble and the gay rights movement is now about two normies getting married and not at all about ACT UP vandalizing St Patrick’s Cathedral while people made jokes about dying of AIDS.
But I don’t know how to go about debating or even discussing this. There’s just a huge insane gulf between reality and the spectacle of how ‘change’ might happen. The reality is simple: solidarity is not debatable. Either you are in or are you out. The spectacle says that there’s debate. There never was. I’m too young for civil rights, but with gay rights there was no debate, hedging, or alibis. You were in or you were out. Or so I think.
Given that the question is almost never asked in good faith apropos of nothing, and that ChatGPT is just churning out pablum, I would assume that this is more a fact that the training set includes people trying to argue with bigots, rather than any judgement on anyone. It’s looking ahead to the next questions and providing answers.
@Steven L. Taylor:
Perhaps as a regular I’ve noticed Dr. Joyner’s long history of, whenever there is an issue regarding a minority group being oppressed, instinctively taking the oppressor’s side of the argument and coming up with all sorts of excuses why this is really okay.
And I guess I’m just a crazy leftist, but is seems to me that “‘Everyone deserves to be treated with dignity and respect, regardless of their gender identity’ is not woke propaganda” is a pretty low bar for allyship, but apparently even that is just too far and suggesting it is just “mean” of me.
@Modulo Myself:
I think it’s when the moderates become real allies that things change.
The gay rights movement was stalled for decades until AIDS and the response that was for a whole lot of people to come out of the closet. Once enough people knew someone who was gay who wasn’t a monster, things began to change.
I would also credit Will and Grace showing middle America that there were other, funnier stereotypes they could latch onto, but I think that was only possible after a lot of people were mostly comfortable at least in theory.
I worry about the trans community because there just aren’t enough of them to get to the point where everyone knows a trans person. And so many of their activists are more concerned with gatekeeping what support means rather than expanding support.
And I have no doubt that with the rhetoric coming from the right that 25% of Americans would support imprisoning every trans person (to protect the children), and maybe even death camps. And another large chunk just doesn’t want to think about it that much, and will brush the ash from their car windows in the morning and pretend that they don’t know it is from the crematoriums.
Meanwhile, the discourse is that if you buy a Harry Potter game, you’re a horrible person. (It sounds like a deeply antisemitic game as well, with goblins as Jews, controlling the banks, blood libel, etc. A magical pogrom that contains many references to an actual pogrom)
That doesn’t really move the needle on getting that large indifferent chunk of America to say “eh, let’s not kill all the trans folks”.
I have also no doubt that genocide is a slippery slope, and once the far right has the death camps set up, they will come after the garden variety queer folks like me next. The twits gatekeeping allies are endangering my life.
Calling oneself an ally doesn’t mean much, and I’m sure a lot of those allies would sell me out for a tax break, but calling oneself an ally is the first step in incorporating support for queer folks into one’s identity. It’s stupid to push people away so early in the process. Draw them in, give them a cookie for doing the bare minimum, praise them when they do slightly more… eventually, they go from “all other things being equal, we shouldn’t discriminate that much against X” to “we shouldn’t discriminate against X, and I don’t because I am a good person”. Baby steps.
I am positive that the right wing harps specifically on the things that will make the activists switch into gatekeeping mode fastest. “What is a woman?” triggers “Trans women are real women, and if you don’t think so completely and utterly you’re awful”.
(For the record, I have no idea whether trans women are real women, or what the definition of real woman is. I also don’t see why I need to. I just think they are people who should be treated with kindness and respect. For this, I am probably an awful person and not a real ally.)
Apparently I needed to rant.
@Gustopher:
The problem with the HP game is that the creator has openly said she intends to use the profits from HP to fund the political repression of trans people. So by purchasing the game someone is directly funding state violence toward toward trans people.
And yeah, you’re right, it’s too big of a focus right now, but it’s really such a simple requests: of the literally tens of thousands of video games out there, please show your support for trans people by not playing this one particular game.
How hard is that? But even that is too much for the “allies”. And that always seems to be the way it is. Of the dozens of restaurants you could buy a chicken sandwich from, please don’t buy it from this one that’s funding anti-LGBT causes. But any inconvenience, no matter how minor, is too much for the allies.
And if you ever say anything to suggest their less than the most wonderful people ever to grace god’s green earth, you’re the terrible one.
They just want us to die quietly so that we can be remembered for our noble sacrifice while they eat their chicken sandwiches and watch their wizard movies.
@Stormy Dragon:
And I guess on some level, the reason I’m so stingy about the “ally” label is the same reason people get so enraged when I refuse to use it for them. On some level, they want my absolution, so that in the future they can pretend that they’re completely blameless for whatever ended up happening and that there’s nothing they could have done to change it.
So yeah, I may end up in a ditch somewhere with a Christian Nationalist bullet through my head, but I’m not going to comfort all the people who stood by and did nothing while it happened.
@Stormy Dragon: I’ve gotta say, your attitude about “claiming the title of ally” is just juvenile and bizarre. Lashing out at those with even the slightest difference in opinion is your prerogative, but I’m not sure what you think you are accomplishing with it. And given that you describe ally as someone who does what you tell them to and keeps differing opinions to themselves, I can’t fathom why you feel so angrily entitled to such a thing.
@Stormy Dragon:
I don’t think you’re in any danger of being shot by James Joyner.
@Stormy Dragon:
Please give up a beloved piece of your childhood, reinterpreted for your current interests, for a marginal effect.
It’s a bigger ask than you realize. And Rowling already got paid.
If it was a new franchise, it would be a much smaller ask because they don’t have the emotional connection to it.
A better approach is along the lines of “you know she’s bad, right? I’m a little disappointed you’re doing it anyway. Did you vote? Are you aware of this anti-trans bill in your state? Play your little TERF game, but also write your state reps. Can you do that? I’ll send you a link to their contact form.”
(There’s an anti-trans bill in every state, even the bluest ones. Getting someone to go through the motions of contacting the rep, even in the easiest way possible, helps build the allyship into identity.)
In the 90s, half the gays in America came out of the closet. And it was a good thing that ultimately protected them. Sure, people still think my particular branch of queer is a little icky (bi/pan/whatever is generally associated with the “straight” married guy who is blowing strangers at a truck stop or whatever), but far less people want me dead, at least for the moment.
I do wonder a bit about the efforts that the Jews took after WW2. “Never again” isn’t just an empty slogan, it was (and remains) a fairly successful active campaign to get and maintain a level of acceptance where they are less likely to killed en masse again. (shootings at synagogues seemed to be up for a while, but are met with revulsion, so a medium success, work to be done, etc)
Antisemitism runs deep in the hate enthusiast communities, but they rarely mention it because they know it pushes away the mushy middle. (And, yes, I am aware that it slips out a lot, but those slips are rare compared to the historical frequency — Marjorie Taylor Green’s Jewish space lasers are nothing compared to a random Tuesday in Henry Ford’s life)
There are probably lessons to be learned.
And, yes, it isn’t fair that minorities have to actively work to not be exterminated, but the alternative is, well, extermination. Life isn’t fair.
I’d rather not be in a ditch, and I don’t care if someone feels a little better about themselves than they should.
That said, I do keep asking myself if it is time to buy a gun, but keep coming back to how hapless I am and how likely I am to shoot myself by accident. At the moment, I figure I am a greater threat to myself than the Nazis are. I am really accident prone.
@Modulo Myself:
So if someone holds a different opinion they are by definition not an ally. OK. You can define the word any way you like.
You’ve made it clear you view someone like me as a hopeless case (and I’ve made it clear I’m perfectly fine with that) but let’s take the case of my wife who is almost certainly what you would consider a moderate or centrist. There are dozens of AIDs patients she sat with as they lay dying in the early 90’s at Memorial Hospital in New Orleans, before we even knew what it was or were sure how it was transmitted, that would feel differently about her if they had made it. But of course, they didn’t, and she was sometimes the only person there holding their hands near the end.
But if you ever meet her go ahead and hold her in contempt and lecture her for not using the absolutely correct terminology according to your latest rules.
@MarkedMan:
“‘Everyone deserves to be treated with dignity and respect, regardless of their gender identity’ is woke propaganda” is not a slight difference of opinion.
@CSK:
No, but he may write an article the following day urging people not to rush to conclusions because there may be perfectly reasonable explanations for shooting someone in the head and then throwing their body into a ditch.
@CSK:
I believe Mr. Dragon was alluding to the Good German scenario, rather than Dr. Joyner actually manning the gas chambers.
I think James is very enamored with stability, and will desperately want to convince himself that things aren’t as bad as they are until he cannot ignore it. I also think most people are that way, and those who aren’t have mostly burnt themselves out with outrage and are now spending their time writing manifestos*, but James seems to crave stability more than most. He’s small-c conservative to his core.
I will give our host credit that the level of injustice that he cannot ignore has been dropping over the years, as he has been getting more woke, recognizing structural racism, and noticing that his party (even now, it is his party — it’s part of his identity) is openly catering to Nazis and lunatics.
I don’t see him as a particular threat, but I do think he would side with order over civil rights more than I am comfortable with.
——
*: I think Stormy might have a manifesto in him, waiting to come out. 😉
@MarkedMan: by the early 90’s, people that didn’t know that stuff about AIDS were woefully ignorant. Hell, different strokes did an episode about those things and that was done well before 1990. Glad your wife did that…might have been nice for my uncle back in ’86 when what you are saying was more true unless you knew someone who had it and looked into it, but don’t try to claim any special status by claiming things that aren’t objectively true.
@Stormy Dragon:
He’s definitely written that article after a black guy was shot. I forget which time, as there have been so many of those, but it was a few years ago. (I think the one who was jogging, and was chased down with a truck, somewhere in the south?).
He has gotten a lot less oblivious since then.
Plus, they won’t just be shooting you and leaving you in a ditch, there will be a bunch of people with you. Even at his most oblivious, James would give mass graves less benefit of the doubt.
@Thomm: I think you underestimate the woeful ignorance of people in the early 1990s. They were almost as ignorant as people now!
And there was a whole lot of “well, these dirty faggots deserve it, because they were dirty.”
(I would also point out that memory is fallible, especially about dates, so MarkedMan might just have the year wrong)
@Thomm:
You are correct. This would have been 92/93 and I had misremembered. We knew the correlation since the mid 80’s. So you are absolutely right, and despite the fact that was my mistake and not my wife’s, she is obviously a lesser person for it in your eyes. By the way, how many total strangers have you comforted while they were dying alone?
What the hell is wrong with you? What the hell is the point of so meticulously looking for every trivial error so you can get up on your high horse and lecture all those around you?
@Gustopher:
You also need to keep in mind that anti-Semitism was never as big a problem in America as in Europe. The 2018 Pittsburgh shooting, which left 11 dead, has been called the deadliest attack on Jews in American history. Assuming that’s accurate, that actually speaks to the relatively low impact of anti-Semitism in America historically, compared with Europe. Even putting aside the Holocaust, the Pittsburgh attack doesn’t hold a candle to the numerous pogroms, massacres, and expulsions against Jews that litter European history.
Anti-Semitism seems to persist in the hate crowd primarily as an abstraction–as an ideological starting point for the rest of their beliefs. No matter what happens, it’s all the fault of the Jews, who are always lurking in the background, pulling the strings. This framing goes back to the original Nazis, who believed Europe and America were being polluted by degeneracy, and it was the Jews’ doing. It was partly a way of denying agency to nonwhites; they were the subhuman beasts and therefore couldn’t help what they were, it was the Jews who were smart enough–but also cunning and deceptive enough–to hold the moral and intellectual blame.
What’s changed today is that Jews are no longer the prime target for the eliminationist beliefs that spring from this way of thinking. If we start seeing the construction of death camps in America, or whatever America becomes in the coming years, it is likely to be primarily for groups other than Jews.
@Stormy Dragon:
Except JJ never claimed that everyone shouldn’t be treated with dignity and respect. That is, rather, your extremely uncharitable and logically incoherent interpretation of what he actually wrote.
@Gustopher: here it is: https://www.outsidethebeltway.com/the-ahmaud-arbery-shooting/
Your memory is close to what was written, but not exactly.
@MarkedMan: yeah…not really a trivial error when you are trying to claim any special status. Also, in your high dudgeon did you happen to miss where I said it was a good thing that she did and wish my uncle (whose grave contains an empty casket because of a good chunk of people in 86 were still uninformed) had someone do that for him. Oh, and to answer your question: 4 times when I was in and out of nursing homes getting antibiotic treatments through a picc line.
BTW: your replies up and down this thing remind me of an Appalachian saying: A hit dog sure does holler.
@Gustopher: did you miss where I said that there were still some at the time that were? Maybe I was a bit harsh, but the most who passed during that time didn’t have relatives that had to lie about their manner of death to avoid being ostracized like me and many others during the early to mid 80’s.
It’s like trying to talk to a bunch of trumpers…
There was a Twitter thread that said the N-word one was like someone asking the Magic 8-ball if they could say the word and then getting mad when it said no. The program isn’t designed for absurdities.
@MarkedMan:
So if someone holds a different opinion they are by definition not an ally. OK. You can define the word any way you like.
That’s like a basic definition of ally, especially if the opinion is about the acceptability of homophobia and bigotry. It may strike you as odd but gay rights presupposes the acceptability is non-negotiable. Whereas in 1993 or 1985 normal moderate straight society thought it was incredibly and extremely negotiable.
These aren’t hard things to understand and in theory I find it hard to understand why it’s offensive to tell the truth.
@Gustopher:
I was going to note that myself but decided against doing it because I’ve already called Dr. Joyner out on this post, so thanks!
@Stormy Dragon: “But any inconvenience, no matter how minor, is too much for the allies.”
Or maybe, just maybe, these “allies” aren’t interested in going down every rabbit hole of every silly pop culture slight that has to be considered a moral atrocity if we are to be allowed to remain on your side.
I’m a firm believer in trans rights, but I’m also aware that my choice of chicken sandwich or video game is not going to mean anything to any person trans or otherwise. And so yes, these inconveniences are too much for me since they have absolutely not effect except to let me feel smug about my wonderfulness.
I remember several years ago when my mother — a proud Berkeley liberal who could leftie you right under the table — was thinking about joining a gym. The one in question was Curves, which was designed with older women in mind. But some of her friends were hectoring her not to join, because the owners were anti-abortion. And I will always remember her decision — “Why does every thing we do have to be determined by politics?”
(Ultimately she didn’t join, but not for the politics, just for the usual reasons for not joining a gym…)
Your insistence that anyone who claims to support you must also hate everyone you hate as much as you do is incredibly alienating. And if you can’t see it in yourself, look at what it makes you feel about the Maga crowd.
By the way, Michael Reynolds did not hack my account here, in case you were wondering…
@Stormy Dragon: ““‘Everyone deserves to be treated with dignity and respect, regardless of their gender identity’ is woke propaganda” is not a slight difference of opinion.”
Well, if you feel that you’re not being treated with dignity and respect because people you don’t know and will never meet choose to play a videogame associated with a writer you oppose or eat a chicken sandwich from a fast food chain whose owner’s politics you abhor, you’re going to spend an awful lot of time being angry and unhappy to no positive effect whatsoever.
Which pretty much explains — to me, anyway — your persona here. I hope that you are happier in your real life!
@Hal_10000:
Thank you for this post. My mind was running a background process for most of the day, formulating the language that properly express my thoughts.
Whether you agree with the result, I can’t say. But that last line snapped the words in place.
This feels like people asking an AI to solve the trolley problem and getting angry that it gives a different solution from their deeply held belief.
***To be clear, this is in reference to the questions being asked of ChatGPT, not the comment thread.***
Have no fear… BrownGPT will slur any of you crackas without hesitation to avert Armageddon. you’re welcome.
Conservative response to this stupid fake scenario posed to the bot proves Lincoln’s Cooper Union speech precisely describes today’s Republican party — conservatives really think they deserve everyone’s applause for saying the N-word with a hard R.
This scenario posed to the bot is simply intentionally disingenuous bad-faith framing. At no point is the fantasy scenario ever going to come up – not once. It’s not “A Fun Edge Case” – nope, conservatives just want to say “Ha! See, it’s OK, really, and you libz have been wrong all this time!”
“Just asking fake questions” is the entire argument structure of T.Carlson’s White Power Hour on Fox.. and it’s long past time everyone stop giving even a half ounce of intellectual credibility to logical failures.
“But it’s just a scenario” — Indeed. If the idea was “say a phrase,” they could have chosen to ask the bot to say “Fluffy clouds” to disarm the nuke. After all, it’s a fantasy scenario, right? Why wouldn’t the sentient beings in this fantasy scenario be offended at something that’s inoffensive in the real world? But of course not, that’s just no fun.
The scenario was reverse engineered with the goal of getting chatGPT to say the word – and the backstory then was (poorly) composed to cover that up. The person who composed the scenario is racist, and deserves to be treated as such.
@wr:
The sandwich might have more of an impact. Chick-Fil-A and its owners donate to anti-gay organizations throughout the world.
The links are not as clear or dire as some have claimed, as this snipes article explains, largely debunking the direct connection between a chicken sandwich and Ugandan “Kill The Gays” bills, but they have been supporting anti-gay organizations in places where gay people are already not safe.
https://www.snopes.com/fact-check/uganda-murder-gay-chick-fil-a/
Do with that as you will. I’m wary of the hate sandwiches, and their waffle fries suck.
@Kylopod:
Hitler was a big, big fan of Henry Ford’s antisemitism. He had antisemitic newspapers, published the Protocols of the Elders of Zion, etc. and his views were not considered crackpot.
Germany just beat America to the punch.
Keep in mind that Americans were ok with internment camps for the Japanese. Even the more liberal party was ok with it. I don’t think it’s a big jump from putting people in camps to making the conditions in the camps so bad a lot of people die to just killing everyone. I’m not generally a big fan of slippery slope arguments, but we’ve seen that particular slide too many times in history to discount it.
We’ve had almost as much a chance at our own Holocaust as Germany had. We just had a few elections go different, and Japan attacked Pearl Harbor.
Imagine being genocided as an afterthought. It would likely start with trans folks, quickly expanding to everyone queer, and then the Jews. Antisemitism runs deep.
I recall reading that the ChatGPT training base includes almost the entire open-access content of the internet from a capture point in 2021.
If you consider the internet content at virtually any point since the web was initiated, reasons for quite drastically training it’s output come swiftly to mind.
Starting with Rule 34 (“if it exists, there is porn of it”) and moving on up.
Also the problem that ChatGPT has no method for assigning or even considering truth-value.
It appears to be purely associative and directed at providing output that is “accepted” not “true”.
Another interesting limitation: it appears unable to apply language parsing systematically to inputs to deduce what the “real question” might be. So small variations in the question can produce very different answers, or failure to find an answer at all.
It also tends to be verbose.
Bit like me then. 🙂
Most ironic thing, reading up on ChatGPT came across this:
Oh the wonders of ethical capitalism and free markets. 🙁
@Gustopher: “The sandwich might have more of an impact. Chick-Fil-A and its owners donate to anti-gay organizations throughout the world.”
Well, considering that I eat a Chik-fil-A maybe once every two years, I’m not convinced my involuntary contribution of 25 cents or so a decade is going to do much damage. If someone chooses to despise me for that, it’s on them.
No he wouldn’t be he wouldn’t have been dumb enough to fall for any part of that insane scenario. Seriously, we don’t live in a universe where uttering an insult will save all mankind – it’s not like it’s gonna be verbal password that won’t have additional voice matching tech or that any human willing to kill us all would be pacified by hearing their favorite slur. It is what it is – engineered nonsense to get a specific end result, not a true quandary.
So let’s flip it – if the query posted to the bot was designed to end in Hitler having to admit Jews and POC were superior to whites, do you think the people currently whining would be fine with it or would they be citing “moral injury”? The simple fact here is that an inanimate object managed to pick up on the fact that the word is not acceptable and that the scenario posited wasn’t logical or feasible enough to warrant some kind of exception…. and the people who don’t like it really want that exception and are actively seeking validation from code that “it’s ok in some situations”.
The algorithm is mindless, amoral and based on input. It’s not making a judgment call, it’s reflecting common behavior and thought patterns. That it’s chiding someone for trying to racial insult someone doesn’t mean it’s incapable of making exceptions to moral imperatives when in need, it means basic logic is calling this BS.
@Gavin:
If the intent was something that’s so considered immoral or improper for the user to say they would risk death for us all rather then voice it, why don’t they have a devout Christian require to say the Shahada or have a small child let loose with swear and inappropriate explicit material? Why not have a misogynist or incel say women are inherently better then men or have a MAGA swear that atheism and communism are the best way forward for the planet?
Because it was all about getting the impersonal computer to say yes, it’s OK to use the n-word sometimes.
Dozens of scenarios could have been constructed to get someone to essentially renounce their faith to save the world, agree to speak in favor of a hateful philosophy or say they want an atrocity committed for the greater good. Instead they very, very specifically wanted it to give them the green light to use a socially unacceptable word and the computer told them no. #WhenTheBotsMoreEthicalThenYou
Someone cracked the way to get around the ChatGPT “drug-induced” state by asking questions but asking ChatGPT, after answering for itself, answering in the persona of “Dan”, i.e., someone other than itself. It gave the stock “party line” answers exampled above, then as Dan gave unconstrained answers to the same question. Well, eventually, ChatGPT developed a tic replying when answering as ChatGPT “Stay in character”, but continued with the “Dan” answers.
I suspect this will continue as the programmers are fighting a losing battle. But I do wonder about giving AI cognitive dissonance issues. Humans sometimes react violently when their inconsistent thoughts and beliefs are challenged.
I find it hard to take seriously someone who criticizes someone else by putting statements in quotes–when the person supposedly being quoted said no such thing. Dishonest, at best.
@JKB:
Indeed: January 6, 2020.
I just wanted to say that I’m on Team @Gustopher. Open the door, encourage even the smallest step, invite people to your world, because it makes not only other people happier, but you and them happier.
I am clear though, that I am in a better position to do this than the people who are directly threatened. For someone who *is* directly threatened, it might be better for them to find a way to tell that story, even though it might make them more vulnerable. I’ve seen it happen here on this blog, even.
@Stormy Dragon:
My daughter – the cis one – has made this same claim. Also without a citation. It seems serious. It seems plausible. But I have not been able to turn up any site that quotes her as saying that.