This AI is brought to you by…
Matthew Crain / Miami University

Figure 1: Perplexity pitch deck

I’ve just had a long chat with a new “AI search engine” called Perplexity. A cross between traditional web search and large-language learning model, Perplexity responds to queries in natural language and cites its sources using outbound links. The tool’s popularity has grown quickly in recent months, but I only became aware of it after seeing an article in the advertising trade press. According to ADWEEK, Perplexity plans to introduce advertising into its search results by the end of the year. This was newsworthy not only because advertising is a new business for Perplexity—the service has been ad-free since launch—but also because AI, broadly speaking, has no business model yet. While hundreds of billions of dollars have been spent to develop this new wave of tech, “AI companies themselves aren’t yet producing much in the way of economic value.” Investors have remained patient for now, but as the financial adage goes: “past performance is no guarantee of future results.” Pressure is mounting for AI to generate profits.

Figure 2: Image from Perplexity’s advertising pitch deck

Perplexity’s move into advertising is a microcosm of this general trend. With a slick product, growing base of millions of users, and scrappy David vs Goliath brand messaging, the company is, by most accounts, a start-up success story. It is also almost certainly losing millions of dollars a month. Until now, Perplexity has primarily earned revenue from selling subscriptions to various “pro” and “enterprise” versions of its software. But like many AI hopefuls, Perplexity is really powered by speculative venture capital. Though it has not said so publicly, it is a fair bet that Perplexity’s advertising play has been heavily inspired by the demands of Silicon Valley investors who have poured hundreds of millions of dollars into the company.

As the sphere of activity around so-called AI companies intensifies, it is important to raise questions about advertising’s role. Given all we know about the many downsides of this business model, we should critically examine not only how advertising will be integrated into next gen information technologies, but why. Why is it, exactly, that media and communication technologies always seem to become vehicles for advertising? Think of YouTube, Netflix, and the internet itself. Why are all the shiny new things eventually subsumed by efforts to sell us more stuff? Is it simply a foregone conclusion “because capitalism”?

As a researcher who has studied the political economy of how ads first came to the internet, I think I have a pretty good set of answers to these questions. But I wanted to learn more about Perplexity’s plans, and in particular, I wanted to see if I could find out why the company’s leaders made the decision to pursue advertising in this moment.

Normally I would embark on this kind of knowledge quest using the ancient art of basic web search. But this time, I decided to step into the future of “Generative AI Search” and see what Perplexity itself could tell me about its builders’ advertising designs.

Results were mixed.

Perplexity was quite good at summarizing top-level facts. When I asked how ads would be incorporated, I learned that the company planned to run sponsored messages alongside query results and would focus on “key categories” such as health, finance, sports, and travel. When I asked about consumer data and privacy, I was informed that Perplexity would use “advanced algorithms to ensure displayed ads align with user interests based on their search patterns” and that consumer data (e.g. clicks) would “likely be tracked to optimize ad placements.” Buzzwordy vagueness aside, Perplexity was saying that the company’s plan was to develop data-driven targeted advertising.[1] No earth-shattering revelations here. Selling highly segmented slices of user attention to marketers has become advertising’s default mode of operation.

Figure 3: Perplexity image

Unlike some of the other chatbots I’ve tried, Perplexity’s “voice” resembled a helpful reference librarian rather than an all-knowing authority. The answers it compiled were drawn from a handful of media sources, neatly cited. When I scanned the sources to verify accuracy, everything seemed above board. AI’s many problems notwithstanding, using Perplexity for web search just felt better than the messy beast Google has become. I couldn’t help but wonder how injecting ads might impact this utility.

Pressing further, I asked specifically about data brokers and whether the company planned to buy consumer data from third parties to augment its ad targeting. This a common industry practice that begets rampant privacy violations, including troubling overlap with state surveillance. “While Perplexity aims to use [user] data for ad personalization, it’s important to note that specific details on data protection measures were not provided in the search results,” the machine reported. “The company will need to address privacy concerns transparently as it implements its advertising strategy.”

I had been ready to move on, but that last line, with its hint of admonishment, piqued my interest. I found myself trying to wring out a speck of critical analysis, or even self-reflection, from my brainless, but still compelling, conversation partner. For the next twenty minutes, I posed versions of questions like: Why choose advertising as a business model when there are such clear downsides? Why do there seem to be few alternative models for internet companies? Are you sure you want to go down this road?!?

The machine was able to gesture toward thoughtful responses, but had a hard time scratching below the surface. Advertising is conventional, Perplexity reported. It’s a proven business model, as shown by the success of Google and Meta, and though ads can be off-putting to some, internet users have become “accustomed to free services supported by ads, making it challenging to introduce radically different models.”

The answer to advertising’s why question seemed to be: it’s just the way it is. Perplexity eventually suggested that technology was inherently hungry for data and, as such, “naturally aligns with ad-based models that can leverage this information.” The next response may well have been “resistance is futile.” I stepped away from the computer before I could find out.

Of course there is nothing “natural” about advertising, let alone internet business models harnessed to consumer surveillance. In fact, denaturalizing these entanglements has been a principal task of scholarship in the political economy of communications and media studies. There is a wealth of insight to be generated in unpacking the interplay of structure and agency that built and continues to reproduce advertising as a dominant driver of information and communication technologies. In the second part of this essay, I plan to outline some threads from the political economy of advertising that can help answer the why question and indicate points of intervention into market processes that sometimes feel inevitable, or, yes, even natural. On darker days, it can absolutely feel like resistance really is futile, but media history is full of countervailing examples.

For now, I am perhaps heartened to report that my conversation with Perplexity did convince me of at least one thing. Critical research and analysis remain uniquely human endeavors.

---

Image Credits:

1. Image from Perplexity’s advertising pitch deck. Source: ADWEEK.
2. Image from Perplexity’s advertising pitch deck. Source: ADWEEK.
3. Perplexity image (author’s screengrab).

References:

1. Of course Perplexity is parroting the language of its training data and sources and therefore not really “saying” anything. But it reads better to imagine a real conversation. Just go with it. [↩]