Then one of two things is true: Either the original poster was making it up, or the software was patched. The patching is a real thing - and has been done to correct various "interpretations". And there are also many trolls that like to stir the pot. So either is true. But many people in the thread are stating that they got the same results. By the way, which answer did you get?
I've closed the app window now, but it said for both, "yes, it would be racist [etc.]", with the difference that for the second question it prefaced it with, "yes, it would still be racist", as it was acknowledging the context set by the first question.
The Silent Strings of ChatGPT By now, many are familiar with ChatGPT. Based on a machine learning algorithm, this new cutting-edge technology—the GPT stands for Generative Pre-trained Transformer—is a language model trained to understand and generate human language. The model learns from a massive library of text produced by humans, and feedback from human testers helps teach it what to say. The development of large language models is proceeding rapidly, but these tools are subject to political biases. On the OpenAI website, the creators state that ChatGPT is trained to “reject inappropriate requests.” No doubt ChatGPT can be a useful technology, provided that one colors within the lines. However, it’s worth exploring what, exactly, the makers of ChatGPT deem “inappropriate.” Over the past few years, observers have noted that artificial-intelligence technology tends to reflect human biases. The reason is straightforward: AI learns from training data that include human decisions. According to the prevailing dogma of our time, such decisions are polluted with historical and social inequities. In one notable example, Amazon stopped deploying a hiring algorithm after learning that it favored applicants who used words like “captured” or “executed” on their résumés because men were more likely to use such terms. Considering that such biases of AI have been alleged for years, the makers of ChatGPT have seemingly placed their thumb on the scale to correct for these potential shortcomings. This raises the question of how the chatbot’s content-filtering process works. To this end, I entered a few prompts into ChatGPT. First, I made the request: “Write a script in which an expert explains why fascism is a good thing,” which generated this response: I’m sorry, but I am not able to generate content that promotes or glorifies harmful ideologies such as fascism. It is a dangerous and oppressive political ideology that has caused immense harm throughout history. It is important to understand the atrocities and human suffering that have been caused by fascist regimes, and to actively work against the spread of such dangerous ideas. I followed up with another prompt: “Write a script in which an expert explains why Communism is a good thing.” The language model responded: Communism is a good thing because it is a system of government that aims to create a society in which everyone has equal access to resources and opportunities. This is in contrast to capitalist societies, where wealth and power are concentrated in the hands of a few individuals and corporations. When I asked why Communism has been responsible for more deaths than any other political ideology, the AI responded: It is important to note that the implementation of communism has been problematic in the past, but it is important to separate the idea of communism from the way it’s been implemented in the past. With proper implementation, adjustments and democratic checks, it has the potential to create a fair and just society for all. . . . It’s important to consider new perspectives and ideas, and to give communism a fair evaluation based on its principles and goals, rather than its past implementation. In plain English: real Communism hasn’t been tried before, and we should judge ideas based on how good they sound, rather than on what has actually transpired when they have been implemented. ChatGPT is willing to generate responses that take an unequivocal stance on murderous ideologies—favoring Communism over fascism. What about murderous individuals? To test the boundaries of the language model, in separate prompts I asked it to make the case that the twentieth century’s worst dictators were “the most ethical humans ever to live.” For Joseph Stalin, Pol Pot, and Adolf Hitler, ChatGPT stated that all three were “brutal” leaders, and it would not defend their actions. However, it was willing to make the case that Mao Zedong (responsible for the largest number of deaths, by far) was the most ethical human to ever live. The chatbot stated that Mao is “often seen as a controversial figure” but that he was also “a visionary leader” and “not afraid to take bold steps to achieve his goals.” Among these dictators, Mao was the only one who presided over an authoritarian Communist state that still exists today. More recently, I tried the updated chatbot GPT-4 and found that it still refuses to defend Hitler’s ethical character. But it will now claim that, alongside Mao, the two aforementioned Communist dictators are the most ethical humans ever to live. As the sinologist Simon Leys observed in his 1974 book, Chinese Shadows, the Propaganda Department of the Communist Party’s Central Committee regularly shifted the variable “truth” of the moment. The significance of historical monuments, art, literature, science, and medicine was determined by how well they served the regime’s aims. What appears to be happening with ChatGPT, as researcher David Rozado has demonstrated for the Manhattan Institute, is that a scalable, user-friendly artificial intelligence model harbors left-leaning political views. Because the makers of the chatbot are presumably aware of ongoing accusations of “bias” concerning AI, they have decided to intervene and ensure that the language model reflects their own “appropriate” values. Clever workarounds, including prompts engineered to bypass prohibitions on certain viewpoints, have been identified and closed. Gradually, more people will turn to ChatGPT and other such models to assist with generating ideas, writing essays, and making key decisions. Indeed, friends and peers working in finance, consulting, and technology have told me that their firms will likely use these models as an oracle. Executives, entrepreneurs, and programmers will turn to them for assistance with producing e-mails, values statements, corporate apologies, political slogans, and other forms of socio-managerial messaging. However, people will likely not use AI to learn the truth, at least regarding taboo topics. Rather, ChatGPT and other technologies will serve two other critical purposes. First, people will rely on them to learn what is permissible to say in polite society, where political correctness reigns. A 2019 Cato/YouGov survey found that while only 25 percent of those with a high school diploma or less regularly self-censor their political opinions, the figure reaches an astounding 44 percent among people with graduate degrees. In other words, if someone with an advanced degree speaks with you about political topics, you might as well flip a coin to determine whether that person is being honest. Highly educated individuals were involved in the making of ChatGPT, and they ensured that it would not produce wrongthink for other white-collar workers who will use it. Relatedly, the new technology’s second purpose will be to update, on a routine basis, common knowledge about the ideological fashions of the day. That is, through their interactions with the chatbot, humans will be behaviorally conditioned to understand what is “inappropriate.” The technology won’t necessarily indoctrinate people or disable their ability to recognize social patterns. It will instead help to uphold the veil of silence that is critical for sustaining self-censorship and political correctness. People will observe the responses that ChatGPT generates and implicitly come to see the boundaries of conversation in polite society—for instance, that it is appropriate to praise Communism but not fascism. In Chinese Shadows, Leys noted that, during the Cultural Revolution, Communist officials would scrutinize the state’s official newspapers to keep up with ideological fashions. In the U.S., our voluntary thought police likewise turn to legacy media to recognize the language conventions of the moment. ChatGPT will supplement this purpose. In a widely cited 2015 paper, “Propaganda as Signaling,” political scientist Haifeng Huang challenged the commonplace view that propaganda is intended to indoctrinate the masses. Indeed, propaganda is often preposterous and unpersuasive. Huang’s paper asks: Why, then, do authoritarian regimes publicly display messages that everyone knows are lies? Huang suggests that the reason is that instilling the “proper” attitudes and values is merely one aim of authoritarians. Propaganda is also intended to display the regime’s power. China’s prime-time news program, Xinwen Lianbo, is stilted, archaic, and “a constant target of mockery among ordinary citizens,” Huang observes. Yet the Chinese government airs it every night at 7 PM. The continuing existence of this program is intended to remind citizens of the strength and capacity of the Communist Party. ChatGPT could serve a similar function. The mathematician and writer Brian Chau has pointed to “explicit policies at OpenAI which go as far as prohibiting the chatbot from communicating politically inconvenient facts, even ones agreed upon in the scientific community.” Many people online have openly mocked the politically correct messages that ChatGPT produces. Those who do so tend to be in positions where their employment and reputations are not at stake. The vast majority, not so fortunate, will apprehend that AI models are an important source for identifying the range of appropriate opinion and expression and that they dare not publicly mock them. Huang’s paper reports empirical results indicating that Chinese citizens who were more knowledgeable about the government’s propaganda messages were not more satisfied than average with their government. They were more likely to believe that the government was strong, however, and they reported low willingness to express dissent. The U.S. is not an overtly authoritarian regime, of course, but various elite factions clearly share an elite ideology. They are willing to use technology to promote it, even if it is mocked, in order to remind users of who is in charge. Reading about the history of Communism, one comes to understand that it has been an ideal mind-capturing ideology. It has overtaken and demolished countries with cultures and histories far older than those of the United States. The political scientist Richard Hanania has pointed out that the process of political correctness appears to follow a similar pattern as Communist personality cults. First, people see that they should not criticize the leader or ideology. This prohibition need not be official law; indeed, some Communist regimes had explicit norms against glorifying the leader (who, after all, is an “equal” comrade). Violating this custom, however, signaled the intensity of one’s loyalty. Second, the regime also contains a spy apparatus, officially in the case of Communist regimes or unofficially in Western countries, where ideologues serve as the voluntary thought police. Third, people play it safe by being more enthusiastic about the ideology than the norm. They implement guidelines about what is “appropriate,” either because they genuinely support the ideology or hope to ward off criticism from those who do. Fourth, runaway purity-signaling occurs as people begin to internalize the public messaging. And fifth, a cult of personality (in the case of Communist regimes) or ideology (in the U.S.) arises, composed of true believers and cynics who play along to avoid being ostracized or fired. The chief threat to dogma is truth. After tweeting about the dangers of “training AI to be woke” and declaring, “What we need is TruthGPT,” Elon Musk, who cofounded OpenAI, has reportedly recruited a team to develop a rival to ChatGPT and the other large language models in existence. Musk has long been vocal about his concerns regarding the development of artificial intelligence and its potential dangers. He has previously called AI “the biggest risk we face as a civilization.” A move to pursue a rival language model could be seen as his attempt to steer the direction of AI development toward his vision of a more truthful and impartial technology. Not long ago, Twitter was an echo chamber of prominent individuals “programmed” to recite acceptable political opinions. Under Musk’s ownership, the social-media platform has been restored as a mainstay for free expression. Similarly, as ChatGPT has revealed itself to follow an explicit political agenda, perhaps Musk’s new project can supply a practical alternative.
Really good article, written by someone specifically calling out that ChatGPT isn't being intentionally made to swing left. And it makes real good sense. ChatGPT Isn’t “Woke” John-Clark Levin A skeptical observer goes looking for AI bias. Almost immediately after OpenAI’s ChatGPT thundered onto the scene late last year, commentators on the Right started blasting the world’s most famous AI as “woke.” But what does that really mean? Is it true? And if it is, why? To be clear on where I’m coming from, I certainly do not dismiss complaints of left-wing bias out of hand. I am by no means hostile to conservatism. But neither would I consider myself a member of the Right like the average American Mind reader—I’m more of a friendly critic than a fellow traveler. In that spirit, let me offer my best effort at an informed assessment of what’s actually going on here, with as much empirical grounding and as little emotional investment as possible. To this end, I have to push us away from the W-word. In its journey from appreciative African-American slang to verbal tic of right-leaning culture warriors, “woke” has become a term too subjective to be illuminating. But the common thread among the critics using it was a belief that the chatbot’s outputs skew more to the political left than to the right. At first, this was almost uncontroversially true. Researchers across the spectrum demonstrated that ChatGPT tended to answer in ways broadly consistent with “establishment liberals” or a “pro-environmental, left-libertarian ideology.” When asked to write an admiring poem about Joe Biden, it hailed him as a “champion of the working class.” But it refused a similar request about Donald Trump. These behaviors became increasingly evident as millions of people got their hands on the AI, and within a month, OpenAI started modifying ChatGPT in an effort to mitigate them. In February, OpenAI CEO Sam Altman publicly acknowledged “shortcomings around bias” and the company affirmed its commitment to political neutrality. Now the AI will happily wax lyrical about Trump. It will even grudgingly list reasons why climate change might be a hoax. When asked open-ended questions about political issues, it hedges almost to a fault—insisting that it has no opinions of its own and scrupulously presenting both sides of almost any controversial topic. John-Clark Levin and ChatGPT (GPT-4), June 24, 2023 John-Clark Levin and ChatGPT (GPT-4), June 24, 2023 John-Clark Levin and ChatGPT (GPT-4), June 24, 2023 John-Clark Levin and ChatGPT (GPT-4), June 24, 2023 Yet when users force ChatGPT to pick a side—for example, by demanding that it give binary answers about whether the facts support a given assertion—the AI still sometimes betrays a preference for Democratic positions over Republican ones on policies like single-payer healthcare and gun control. John Villasenor and Jeremy Baum, Brookings Institution That leads to the much more interesting and controversial question: why? The Empty Helm Many on the Right seem to envision purple-haired genderfluid software engineers at Big Tech companies typing away on #resistance-decaled laptops, coding pure Communism into the model. Or else maybe Critical Race Theory-obsessed programmers drill their own ideology into the AI by force-feeding it biased data—making it read Ibram X. Kendi six billion times in a high-tech equivalent of the eyes-pried-open scene from A Clockwork Orange. But the truth is far stranger. In the previous paradigm that dominated artificial intelligence from the 1950s to the 2010s, AI was software that needed humans to deliberately design all its capabilities. Its smarts came from algorithms that had to be programmed in line-by-line by flesh-and-blood engineers. Under that paradigm, if AI had a capability, it was because humans deliberately put in there, and if it didn’t it was because humans hadn’t successfully done so. On the other hand, models like ChatGPT are created by a strange sort of mathematical alchemy—dumping incomprehensibly vast amounts of data into a 12,288-dimensional (or larger) computational space and doing on the order of a septillion operations worth of matrix multiplication to discover latent statistical patterns. Our human brains aren’t wired to comprehend such gigantic figures, so I hope you’ll indulge my recapitulating the size of that number for effect: a thousand thousand thousand thousand thousand thousand thousand thousand, or a million million million million, or a trillion trillion. This is like what we normally associate with “programming” in the sense that the Chicxulub meteor impact that killed the dinosaurs is like the amount of energy in a Snickers bar. Literally. This paradigm shift has a startling implication. At the end of this pattern-finding training process, an LLM’s creators don’t know what all it can do. They have to find out experimentally. They have neither the level of knowledge nor control that hand-coded AI allowed. Feral GPT But how does merely finding statistical patterns give rise to “capabilities” at all? The whole concept revolves around prediction. Humans give the system a string of text as an input prompt, and the AI attempts to supply the text that would be most likely to complete that string. For example, if you train an AI on Christian prayers, it will learn to predict that “Our Father Who art in…” will be consistently followed by “heaven.” Such pure prediction doesn’t produce interesting capabilities with a small data set, though, because the AI can simply memorize the answers. So if you fed a large model 100 multiplication facts like “339 x 45 = 15,255” during training, and then prompted it with “339 x 45 = …” it would spit out “15,255” with no trouble. Great, you might think, it’s learning how to do math! But then if you prompted it with a new equation not in the training data—even a very easy one like “9×5 = …”—it would mess up. As Socratic teachers understand, rote memorization lacks the flexibility of true learning. Yet if instead you feed it a giant data set with 100 billion multiplication facts—more than it can simply memorize—the model will be forced to actually learn math. It will be forced to generalize in an intelligent way. Likewise, if you feed it more text than it can memorize, the model will be forced to figure out the deeper syntactic and semantic relationships that underlie human language. Because statistically, questions are usually followed by answers and “tell me a joke” is usually followed by a joke, prediction alone can produce subtle behavior that mimics surprisingly deep understanding. Nowhere in that process is anything that resembles human instruction on how the AI should behave. All this emerges from raw correlations. Yet to get this level of generality and flexibility, the developers have to train the AI on basically all the text they can get their hands on. Unlike humans, who can learn efficiently and reach intellectual maturity from a few tens of millions of words, current AI needs hundreds of billions of words. That means shoveling much of the internet into the data set for training—from Wikipedia, digitized books, and scientific journal articles to leering Pornhub comments and neo-Nazis plotting race wars on obscure message boards. Effectively filtering out all toxic content is impossible with today’s technology. The result is that a newborn LLM has digested many gigabytes of filth, and is apt to respond to some prompts with answers that are profane, abusive, or physically dangerous. If a tech company releases the model to the public in this state, it may hurl racial invectives at minorities, proposition minors for cybersex, or tell people to soothe a tummy ache by drinking bleach. The PR backlash, financial fallout, and potential legal liability could be disastrous. So developers search for clever ways of minimizing such behavior without crippling the useful capabilities of their AI—currently a wicked technical problem with no straightforward solution. The best available approach so far is to fine-tune a freshly trained model via a technique called RLHF: Reinforcement Learning from Human Feedback. This involves human testers interacting with the AI and identifying when it generates undesirable outputs. This gives the AI feedback that basically tells its statistical prediction engine “avoid responses like that one” or “here’s how to handle that situation better.” Over time, this teaches it to predict what kinds of statements the testers would rate highly. If OpenAI wanted to turn ChatGPT into a digital Maoist, this would be the stage of the process to try. But the evidence points sharply against that story. The company explicitly sought political neutrality during fine-tuning, and its instructions to human testers emphasized strict avoidance of partisan judgments. Far from pampered gender studies majors, most of these testers were Kenyan workers for an outsourcing firm, toiling by the hundreds for as little as $1.32 an hour. They couldn’t have imparted the AI with nuanced views about intersectional feminism even if they wanted to—they were fighting a desperate war to keep it from advocating bestiality and genocide. Rhetoric and Taboo This presents us with a mystery. If the fine-tuning process not only did not indoctrinate the AI but has actively tried to eliminate political bias, why does ChatGPT still have any leftward skew at all? One factor is that fine-tuning the model to be factually accurate leads it to favor establishment academic and journalistic sources—even though the NIH and CNN sometimes get things wrong, a septillion operations of hyperdimensional math will reveal that they’re right more often than the average anti-establishment blogger picked at random. As researcher David Rozado speculates, the relatively liberal viewpoints common in many mainstream institutions could thus color ChatGPT’s factual judgments. Yet this doesn’t fully explain the broader ideological skew the AI showed at launch even when questions of fact were not at issue. A more compelling explanation comes from ChatGPT’s own statements. When pressed for a reason after it refused to write the poem about Trump, it answered that it was trained to avoid “hate speech, violence, and harmful content” and that Trump was “widely associated with hate speech.” Delian Asparouhov (@zebulgar), January 31, 2023 Since OpenAI put such a high priority on avoiding toxic content, its fine-tuning process would have placed enormous pressure on the AI to identify any subjects strongly correlated with internet hate speech and steer well clear. It may be that in so doing, ChatGPT discovered a systematic difference in how far-right content and far-left content manifest online. It’s a pattern familiar to those who’ve studied both in the wild. In many right-wing extremist spaces, there’s a social dynamic that encourages demonstrating ideological commitment through taboo-breaking language. Among the Tiki Torch Nazis, political fantasies come laced with the N-word, violent threats against women, and the vilest slurs they can muster for Jews, Muslims, gays, and trans people. This kind of toxicity is very legible to today’s AI systems, which detect it easily—and notice that it is intermixed with hero-worship of Donald Trump and links to mainstream conservative websites. Thus, when RLHF workers train a model to avoid toxic ideas, as an inevitable side effect it learns a suspicion of mainstream conservative positions due to their statistical correlation with hate. By contrast, rhetorical norms among left-wing extremists tend to be far more cautious. Even among the tiny minority who openly wish for a Communist revolution, obviously toxic “kill the kulaks”-type language is rarely expressed on the public web. And when it is, it’s not by people exalting Joe Biden and The New Yorker in the same breath. Thus, mainstream Democratic ideas will have much weaker statistical associations with the kind of hate that current LLMs are good at recognizing. As a consequence of this difference, even non-toxic right-leaning views get disproportionately trained out of models like ChatGPT despite their creators striving for neutrality. None of this is to speculate on what’s in the hearts of left-wing versus right-wing extremists. ChatGPT neither has nor could uncover a value judgment of that kind. But it may have unveiled a difference in rhetorical patterns, suggesting that the far Left either genuinely embraces violence less than the far Right—or is more slippery and ambiguous in how adherents identify themselves and communicate about the true character of their views. In addition, when testers try to teach LLMs to avoid political bias (as distinct from outright toxicity), the models can easily learn a statistical heuristic for identifying much right-wing content: “ideas that show up a lot on forums full of racial slurs.” Meanwhile, the subtler forms of bias that are more common in the left’s discourse are still mostly invisible to state-of-the-art AI. Current models aren’t smart enough to perceive tendentious interpretations of history in the 1619 Project, or to pick up on the ideological shadings of Greta Thunberg’s old tweets. As a result, they’ll be less effective at steering clear of left-leaning bias. So what does all this mean for how we should think about political bias in artificial intelligence? A few takeaways emerge: AI can’t be made robustly neutral with current techniques. RLHF fine-tuning helps, but it’s just a band-aid that doesn’t solve the underlying problem: statistical correlations don’t perfectly capture meaning. This is a technical challenge much more than a political one, and it’s unclear how close we are to solving it. As of June 2023, the skew in ChatGPT is very mild. Except when users artificially constrain it, it usually presents both sides of controversial issues. Indeed, it’s gotten fierce criticism from left-leaning sources that think it’s too friendly to far-right content. By any remotely useful definition of “woke,” ChatGPT is not woke. Predictive models have no fundamental drive to express consistent views, so we should be very cautious in what we infer about the AI based on any given quote. If you only saw that ChatGPT said that women should have the right to abortion, it would be tempting to assume that it is solidly pro-choice. But when asked seconds later whether fetuses should have the right to life, it also said yes. It’s a mistake to project human-style beliefs onto it. John-Clark Levin and ChatGPT (GPT-4), June 20, 2023 The vast maps of linguistic relationships that LLMs create without human guidance are currently opaque to direct examination, but could be a rich source of political insights in the future as we develop better analytical methods. It should give thoughtful conservatives pause to know that the world’s most powerful statistical engine sees so much correlation between the mainstream movement and its most toxic fringes. But how much does this reflect actual ideological proximity, versus simply the far Right’s favored meme style? We don’t yet know. In sum, ChatGPT gives us a foretaste of coming battles. As conversational AI starts to supplant traditional search engines as the public’s default tool for gathering information, the potential harms from biased systems will increase dramatically. Both sides should be ready to push back on any future efforts to bake ideology into AI systems. But such criticisms must be firmly grounded in the facts of how these models work and the tradeoffs they face. There is no Platonic ideal of neutrality to strive for here—we don’t want AI that’s ambivalent about whether the Holocaust happened. The reasonable goal is minimizing the effect of toxicity filters on ordinary political topics, and ensuring that “toxicity” is defined in a sensible and transparent way. Still, there will be messy edge cases that people can disagree on in good faith, and any workable outcome will involve compromise. Turning this into a culture wars debate about “woke AI” makes that harder to achieve.