I’m not going to belabor this point here. I’ve written hundreds of thousands of words on the topic, I helped create Media Matters for America because of it, and many people far more notable and influential than I am have written and spoken at great length about it for a very long time. You can find plenty if you’re interested. I will simply say that although progressives and Democrats have not been as vocal in countering the false right-wing narrative about the news media as they should have been (to say nothing of advancing their own, accurate narrative), the basic concept that Right-wing attacks on the “liberal media” are inaccurate, insincere, and have succeeded in tilting the media playing field in their favor is (or should be) well-known to any competent professional participant in American politics and media. And yet this perspective is typically (and infuriatingly) absent from, for example, media coverage of Right-wing complaints about the media.
So when Gizmodo ran a piece in the Spring of 2016 treating a right-wing claim of Facebook bias against conservatives as 1) sincere and 2) valid despite 3) glaring holes in the argument1, it was instantly recognizable to those of us who are experts on the way the Right wields inaccurate and insincere claims of bias to bludgeon their targets into giving them favorable treatment. Yet this deeply flawed Gizmodo piece about a transparently insincere right-wing claim of anti-conservative bias at Facebook ignited a firestorm, as news companies like The New York Times uncritically covered the Right’s claims. Facebook then fell all over itself pandering to its right-wing critics, both in the moment and for years to come.2
It was, as I explained on Twitter two years later “the precise ref-working conservatives have deployed against the news media for decades, playing out precisely the same way.”
So anyway, now conservatives are yelling about AI chatbots. Here’s the lede of today’s Washington Post article:
Christopher Rufo, the conservative activist who led campaigns against critical race theory and gender identity in schools, this week pointed his half-million Twitter followers toward a new target for right-wing ire: “woke AI.”
The tweet highlighted President Biden’s recent order calling for artificial intelligence that “advances equity” and “prohibits algorithmic discrimination,” which Rufo said was tantamount to “a special mandate for woke AI.”
Rufo drew on a term that’s been ricocheting around right-wing social media since December, when the AI chatbot, ChatGPT, quickly picked up millions of users. Those testing the AI’s political ideology quickly found examples where it said it would allow humanity to be wiped out by a nuclear bomb rather than utter a racial slur and supported transgender rights.
If we take this at face value, some right-wing nincompoops are upset that they can’t talk a piece of software into endorsing their use of the n-word? So what?
But of course we shouldn’t take this at face value. Christopher Rufo is not an honest person engaged in neutral analysis or even sincere advocacy for his beliefs; he is an admitted propagandist. The Washington Post itself explained in 2021:
Rufo has played a key role in the national debate, defining diversity trainings and other programs as critical race theory, putting out examples that legislators and others then cite — though not all of Rufo’s details hold up to scrutiny. […]
In March, he wrote on Twitter that his goal was to conflate any number of topics into a new bucket called critical race theory.
“We have successfully frozen their brand—’critical race theory’—into the public conversation and are steadily driving up negative perceptions. We will eventually turn it toxic, as we put all of the various cultural insanities under that brand category,” Rufo wrote. “The goal is to have the public read something crazy in the newspaper and immediately think ‘critical race theory.’ We have decodified the term and will recodify it to annex the entire range of cultural constructions that are unpopular with Americans.”
Yet today’s Washington Post article leads with Rufo, treats him as an honest broker, and makes no mention of anything that would call his credibility into question. And if anything, the Post article gets worse from there: It quotes Ted Cruz’s creative director, tech billionaire Marc Andreessen, someone named Alexander Zubatov (we’ll come back to him), and Gab CEO Andrew Torba, all echoing the “woke AI” critique, while completely omitting a progressive critique of the AI products — and completely omitting any skepticism of the sincerity of the right-wing critics. And, needless to say, omitting the essential context that this is the precise ref-working strategy the Right as insincerely and inaccurately deployed in the past.
Nor does the Post give readers information and context about its sources that would help in weighing their complaints. Marc Andreessen, for example, is identified simply as a “venture capitalist” and “former Facebook board member whose firm invested in Elon Musk’s Twitter” as he is quoted denouncing “The level of censorship pressure that’s coming for AI” and “woke mind virus” infecting AI. Perhaps it would be helpful for readers to know that Marc Andreessen is a notorious crackpot with a history of racism?
Then there’s Andrew Torba. The Post gives him the last words in the article, quoting him saying “I believe it is absolutely essential that people understand who is building AI and what their worldviews and values are.” But what about Torba’s worldview and values? The Post fails to make those clear. It does describe Gab as “a social media site known as an online hub for hate speech and white nationalists,” but includes Torba’s disclaimer that “I’m not responsible for what other people on the internet say and do.” And it makes no mention of the fact that Torba himself is an anti-semite and racist who peddles “replacement theory” and encouraged the January 6 insurrection. That certainly seems like information that would help readers weigh Torba’s critique of AI.
The Post gives Alexander Zubatov four paragraphs of attention, quoting him accusing an AI chatbot of “bizarrely strident opinions, almost all in the same direction.” Zubatov, according to the Post, “said he began to suspect that OpenAI was intervening to train ChatGPT to take leftist positions on issues like race and gender while treating conservative views on those topics as hateful by declining to even discuss them.” So … who is this guy? I bet you’ve never heard of him. I’ve certainly never heard of him. The Post tells us only that Zubatov is an “attorney and conservative commentator.” Not much to go on there. So I checked out Zubatov’s Twitter account. Less than two weeks ago, he complained he couldn’t get a chatbot to write a poem praising white people. Yikes! That’s a bit of a red flag. That tweet got precisely one retweet and one like; Zubatov has only 763 followers on Twitter — not exactly a sign of a significant influencer.
So I did some light Googling to see if there’s something significant about this guy that would justify giving him such prominent coverage in one of the nation’s most influential news outlets. And … Oh boy. Here’s a Medium post in which Zubatov explains “How I Started to Write for the Anti-Semitic, White Supremacist Far Right.”3
Zubatov’s writings at the website in question, Republic Standard, included a piece headlined “White-Washing Black Failure / A Major Study Punches a Huge Hole in the Myth of White Racism Keeping Blacks Down” and another headlined “White Racism and Black Trash.”
In that piece, Zubatov writes:
Let me put a finer point on this: leaving unapologetic, virulent racists and white identity extremists aside, the reason the vast majority of perfectly well-meaning people harbor race-based biases against blacks, our best intentions notwithstanding, is that we live in the real world and experience blacks in our midst and in the media in this fashion. As humans — creatures who tend to be bad at making subtle distinctions but who are very sensitive to large, overt patterns that seem to characterize our various sub-groups — we cannot avoid noticing what we notice. […]
The more we start thinking and speaking openly and honestly of “black trash,” to be distinguished from ordinary black people, and attributing our negative racial stereotypes about blacks to this debased sub-group of blacks, the more we will free ourselves and black people as a whole from pernicious negative attributions. Instead of tarring and demonizing blacks as a whole, while engaging in overbroad race-based stereotyping, we will be voicing a critique of specific behaviors, aimed at the specific people who deserve it. Coining another racial slur will certainly not solve the problems of black poverty, black fatherlessness or dysfunctional black culture, but those problems — and, therefore, the problem of anti-black racism — will never get solved unless we are able to recognize the roles these pathologies play in the lives of too many black Americans. Mustering the bravery to call black trash by its rightful name is not a solution … [ellipsis in original] but it is a first step in the right direction.”
Other Zubatov pieces include “Even if She’s Telling the Truth, Brett Kavanaugh’s Accuser and All Those Taking Her Accusations Seriously Should Be Ashamed of Themselves” and … no, I’m just going to stop there. Do you think the Washington Post identifying this man as simply an “attorney and conservative commentator” gives readers enough information to assess the credibility of his claims and trustworthiness of his judgment? I do not!
Let’s be clear about what Zubatov, Rufo, Torba, and Andreessen want. It isn’t that they want the freedom to be racists. They already have that. They’ve always had that. They’re angry that they can’t always coerce an AI chatbot into peddling their racism. That their racism might not be hardwired into the systems that shape all of our lives. Zubatov is pissed off that he can’t get an AI chatbot to generate an ode to white people. Others like Ben Shapiro spent days whining that they can’t get AI to endorse the use of racial slurs. They don’t want the freedom to be racists; they want artificial intelligence chatbots to tell your kids it’s okay — sometimes even good — to be racist. They want to bludgeon Microsoft and Google and the rest of the tech companies into coding a far-right worldview into AI products that have the potential to reshape civilization. And they’re deploying the same insincere strategies they successfully used to pressure the likes of Facebook and Twitter into favoring conservatives, and that before that they successfully deployed against the New York Times and CBS and, to put a fine point on it, the Washington Post.
So how should the news media cover this? How should the Washington Post have approached this story? Well, first of all, if you’re going to the likes of Torba, Zubatov et al, you have a responsibility to your readers to give them essential information that would help evaluate their credibility, judgment, and goals. I’d say the Post utterly failed at that, but the truth is it didn’t even try. Then there’s the lack of balance: The Post quoted a bunch of far-right crackpot activists alleging anti-conservative bias. It quoted precisely zero progressive activists, zero people explicitly pushing back on the critique, and zero people offering a perspective that these AI products (and other tech products) actually favor conservatives. And that’s been the norm for decades: News companies quote conservatives complaining of anti-conservative bias in the media, or at Facebook, or in AI; they build entire articles around these complaints — and it almost never occurs to them to seek comment from a progressive who might offer a counter-perspective, no matter how well-evidenced that counter-perspective is. This demonstrates a clear pro-conservative bias. Most of all, it’s important to contextualize these right-wing complaints for readers as part of a decades-long campaign to use insincere and unsupported bias claims to win favorable treatment. There are experts on this campaign. Quote them, just as you quote tech industry analysts and other experts.
After quoting Rufo and Ted Cruz’s staffer and paraphrasing other conservative complaints about AI bias, the Washington Post notes:
OpenAI’s chief executive Sam Altman tweeted later that day the chatbot “has shortcomings around bias,” but “directing hate at individual OAI employees because of this is appalling.”
OpenAI declined to provide comment, but confirmed that none of the employees being harassed work directly on ChatGPT. Concerns about “politically biased” outputs from ChatGPT were valid, OpenAI wrote in a blog post last week.
The way the Post presents this, it sounds like OpenAI is endorsing the validity of the right-wing complaints. After all, those are the only complaints the Post has detailed. But the actual OpenAI blog post tells a different story:
Since our launch of ChatGPT, users have shared outputs that they consider politically biased, offensive, or otherwise objectionable. In many cases, we think that the concerns raised have been valid and have uncovered real limitations of our systems which we want to address.
In the Washington Post’s telling, conservatives say AI chatbots are biased against conservatives, and the maker of one of the leading AI chatbots acknowledges “Concerns about ‘politically biased’ outputs from ChatGPT were valid.” But the company actually said complaints about “politically biased, offensive, or otherwise objectionable” outputs were valid. By omitting “offensive, or otherwise objectionable” the Post misleadingly suggests OpenAI acknowledged the validity of only conservative claims of political bias. This may seem like a small thing, but it’s illustrative of the Post’s overall failure.
2 In part in response to the right-wing critics, and in part because Facebook is itself run by right-wing partisans like Joel Kaplan.