BuzzFeed’s AI-Produced Content Experiment Is a Glimpse Into a Bleak Future

The internet media company BuzzFeed, which produces short articles, quizzes and lists such as “18 Ridiculously Cosy Clothes From Amazon That Have Been Getting Us Through The January Blues” was struggling. After going public in December 2021 with a $1.5 billion valuation, the company’s stock dropped precipitously, losing 80 percent of its value by the end of 2022. But on January 26, 2023, BuzzFeed CEO Jonah Peretti released a memo describing how the company would integrate OpenAI’s ChatGPT large language model into its production processes. As a result, Buzzfeed’s stock saw a rapid 150 percent increase in valuation.

OpenAI, which was founded as a nonprofit in 2015, became in 2019 a de facto Microsoft subsidiary in return for access to the tech giant’s vast computing power. OpenAI is a leader in the form of artificial intelligence (AI) known as large language models (LLMs). An LLM is a statistical model of a language, created by machine learning, which contains within it the probability that any given sequence of letters occurs. LLMs can generate text output in the form of probabilistic projections from an input prompt. Given the prompt “It is rainy and…” an LLM might output “dark” or “windy.” Sophisticated LLMs like ChatGPT can do much more, from generating passable graduate-level essays to working software code.

The exuberant reaction of the market to Peretti’s memo provides an example of the tech industry’s embrace of buzzwords and bandwagonism. But it also provides an opportunity for thinking about the capitalist context in which AI is produced and used, and the vacuity of industry-driven discourse on AI’s social implications and ethical dimensions.

The Point Is Profit

Since around 2015, the U.S. tech giants which lead AI research and development — Alphabet (Google), Meta (Facebook), Microsoft, Amazon and Apple — have made a concerted effort to portray their AI as “ethical” and to paint bright pictures of the future where AI improves life for everyone. Academics have joined in too, offering theories and models to make AI more ethical.

“Ethical AI” discourse tends to ignore AI’s industrial context. In the simplest terms, this means recognizing that AI is, by and large, a commodity. This means that it is produced with the intention of being sold to generate a profit. Because not all applications of AI are profitable, this simple insight has hefty ramifications for thinking about AI, its social and ethical implications and possible futures. AI’s industrial context is left out of industry and industry-supporting academic discourse for this very reason. AI thus tends to be conceived of as a discrete technology whose ethical and social implications can be dealt with on a purely technical level, and certainly without engaging in historical and systemic analysis of the capitalist economy.

However, the driving force of capitalism’s dynamism is not technology, but competition between capitalist firms, which drives them toward “constantly revolutionising the instruments of production,” as Karl Marx and Frederick Engels observed. Automation, the use of technology to increase the productivity of labor, is also driven by competition. If your company can use ChatGPT to produce three high-impact articles using the same amount of labor time it takes my company to produce one by hand, your company will have a decided competitive advantage.

What BuzzFeed’s Memo Tells Us About AI

ChatGPT presents truly startling capacities for the algorithmic manipulation of language. It represents a significant scientific and engineering triumph which may end up telling us a lot about how human cognition works and the differences and similarities between how humans and machines process information. Peretti’s memo, which stoked the fires of investor excitement, described a mundane application, moving “AI inspired content … from an R&D stage to part of our core business, enhancing the quiz experience, informing our brainstorming, and personalizing our content for our audience.” In sum, to “maximize the creativity of our writers, producers, and creators.” In other words, increasing the productivity of labor engaged in creating clickbait.

The narrow focus on increasing productivity leads to contradictory and vacuous discourse, as the primary goal of capitalist production — the capture of value — must remain unspoken. The main webpage of Google AI proclaims a commitment to AI which is “socially beneficial, fair, accountable, and works for everyone.” It does not mention that such niceties are only possible if Google can continue to generate a revenue stream. It especially does not mention that Google’s revenue stream (and that of the other big tech firms), as many studies have shown, is premised precisely on unaccountability, a lack of concern for social benefit and a very vague interpretation of fairness based on the ceaseless and aggressive acquisition of personal data — what Shoshana Zuboff calls “surveillance capitalism.”

This sort of vacuity manifests in the BuzzFeed memo when Peretti describes the company as aspiring to be the “premier platform for creative people to do their most inspired work, reach the most people, and live their best lives” only to follow this up, in the next paragraph, with the announcement that BuzzFeed will “expand beyond AI-powered curation (feeds), to AI-powered creation (content).” In other words, from using AI to recommend hand-written articles to a user, BuzzFeed will use AI to generate those articles. Peretti goes on to describe a “new era of creativity, where creative humans like us play a key role providing the ideas, cultural currency, inspired prompts, IP, and formats that come to life using the newest technologies.” The idea is not to replace workers entirely — says the company which laid off 12 percent of its workforce in 2022 — but to have them feed ChatGPT with novelty so it can keep churning out clickbait. Here, the appeal of the creator economy — its perceived authenticity — is to be achieved, as much as possible, without creators.

But this is not the only contradiction on hand. A more blatant one manifests in the “ethical AI” discourse boosted by industry, and taken up by academia and governments. Google asserts that “values-based AI is good for your business.” One might reasonably ask: which values? Are they Confucian, Christian, socialist, liberal, or something else? The answer, of course, is the values that do not conflict with the imperatives of capitalist production. Ethical AI is a dead-end endeavor, set up to fail by its basic terms because it refuses to recognize the basic dynamics of capitalist production driving the incorporation of AI into the labor process and simply assumes that more automation will be better for everyone.

The BuzzFeed memo makes it clear that the stakes of introducing AI into business processes are economic, rather than the ethical. Peretti describes a need to “fiercely focus on delivering strong value for our partners so that they continue to spend with us during the recession.” Making clear that increasing the productivity of labor is an explicit goal, he notes that, “In tough economic times, we need to … save every penny of costs.” The dubious nature of the AI industry’s ethical claims does not require a radical leftist to recognize it. The neoliberal economist Milton Friedman noted, back in 1970, that the only “social responsibility” of business is “to increase its profits.”

The BuzzFeed memo might be mere hype. The strategic AI plan might fail to pan out. Or the company might rise to new heights of profitability on the basis of AI-generated clickbait. In either case, the memo paints a bleak image of the capitalist vision for AI: trivial applications designed to wring ever more dregs of productivity from labor, peddled alongside vacuous discourse of social benefits and ethical betterment.