Google ruins Search. OpenAI: hold my beer

Why OpenAI’s publishers program signals a big shift for AI business models

It’s no secret that Google destroyed Search by crowding the top results with SEO rubbish and advertising content. A leaked pitch deck from OpenAI confirms it’s about to do exactly the same thing with ChatGPT.

According to a deck obtained by Adweek, OpenAI has been spruiking an initiative called the Preferred Publishers Program, which would allow media companies to pay to receive preferential treatment in ChatGPT’s conversations with users.

For the right price, companies can now enjoy a “richer brand experience” in chat including in-line product inserts, clickable links, branded buttons and anything else OpenAI thinks it might be able to sell to advertisers. The result? Poorer quality text, more branded content and a steady stream of garbage flooding our screens.

But what does this move tell us about the future political economy of AI? Google had an incredibly popular and useful digital tool that was famously unprofitable until it discovered advertising. Today, Google and Meta remain largely advertising platforms; Meta still makes 97.5% of its revenue this way.

OpenAI started off down a different path. It offered users subscriptions for its premium service and charged businesses to access its API. Its annualised revenue recently hit $2 billion and its growth rate continues to soar.

OpenAI has been rumoured to be launching a Search competitor, which will be integrated within ChatGPT.  Most likely this will enable chat conversations to search the Internet presumably powered by Microsoft’s Bing. It makes sense that as it attempts to turn its own product into Search it would also turn to Google’s primary revenue model of advertising.

Previous publisher partnerships including OpenAI’s deals with Associated Press, Axel Springer, Le Monde, Financial Times and others were about gaining access to training data and are a net cost for OpenAI. These new deals look like an attempt to build a new revenue stream through advertising.

These new partnerships would be a pivot from a Spotify/Netflix style platform to a Google model. Instead of individuals paying to use the service, we will become the product as OpenAI allows advertisers to distort results. Perhaps the final outcome will be a hybrid model with diverse revenue streams, including potential future infrastructural investments in hardware and energy facilities.

But will this be a path to profitability? Although revenue continues to climb, so do OpenAI’s training and infrastructure costs. Sam Altman has confirmed that OpenAI remains lossmaking and its expenses are likely to increase as it trains ever-more sophisticated models requiring tens of billions of dollars.

OpenAI’s generative text also has an odious impact on the broader online ecosystem. AI-generated content continues to poison the Internet with low-quality regurgitated text. There are currently no rules to prevent the general degradation of online spaces with enormous amounts of AI-generated trash.

According to Altman, 92 per cent of Fortune 500 companies are currently using OpenAI products. However, the rise of generative AI has also spawned serious counter-movements and a rejection of the use of AI in creative content by some fans. Many people are reminiscing about earlier eras of the Internet and looking for possible alternatives today.

OpenAI’s business model continues to evolve. While most signs point towards its ongoing rise, there is an underlying dissatisfaction with how the Internet has evolved, which highlights deep contradictions between OpenAI’s PR campaigns and how it actually intends to build its revenue.  The Preferred Publishers Program along with the move into Search might be a pivot towards an older platform revenue model, one that could signal a broader shift in AI business models for leading firms.