Shrimp Jesus and the AI ad invasion - Meta Platforms CEO Mark Zuckerberg — Meta Platforms CEO Mark Zuckerberg

Shrimp Jesus until now has been one of the internet’s strangest artificial intelligence fever dreams. The likeness of the Christian figurehead fused with crustaceans started going viral on Facebook last year, epitomising the rise of “AI slop” or images generated by tools like ChatGPT, graphics that millions of social media users have inexplicably liked and shared.

Mark Zuckerberg has lapped it up. In recent months, he’s given a string of podcast interviews describing social feeds that will be filled with AI content and AI friends. And fuelling this synthetic future will be AI ads, according to The Wall Street Journal, which reported on Monday (paywall) that Meta Platforms will offer tools for advertisers to create ads with artificial intelligence, corroborating Zuckerberg’s own hints on the matter. One day, Shrimp Jesus might try to pitch you on a local seafood restaurant.

This has set off palpitations at ad agencies. Who needs the expertise of today’s Mad Men if Meta’s tools not only target users but also do all the creative work of building adverts from scratch? Shares in WPP, a British advertising giant, fell 3% in London trading on Monday after the Journal published its report and remained flat on Tuesday in a rising stock market.

But the consequences aren’t so cut and dried. Most of Meta’s advertisers are small and medium-sized businesses spending around US$5 to $20 to shop their wares on Facebook or Instagram, compared with larger firms who will pay WPP or other agencies thousands of dollars for a campaign. For the smaller companies, Meta’s AI ads could be an opportunity to more effectively reach new customers, cheaply. Here’s how Zuckerberg — whose company derives 97% of all its revenue from advertising — described the process in an interview with Stratechery’s Ben Thompson.

“We’re going to get to a point where you’re a business, you come to us, you tell us what your objective is, you connect to your bank account, you don’t need any creative, you don’t need any targeting demographic, you don’t need any measurement, except to be able to read the results that we spit out.”

Consequences

Simple, right? And Meta won’t just streamline the advertising process for small businesses. It will offer a powerful new form of personalisation, based on the reams of personal data it collects. Meta’s AI, for instance, will generate an ad for a car driving up a mountain if that’s where the user is based, or through a city if they live in an urban area, according to the Journal.

Sure, these tools likely won’t come up with “Just Do It” or the “Diet Coke break” ads, and there’s a good chance that hallucination-prone AI tools will make the odd gaffe that embarrasses some brands. But given how willingly social media users have accepted AI content already, there’s a good chance these new ads will make money. AI-generated video has also improved dramatically in the past few months, with users of Google’s new Veo 3 tool posting astonishing clips that depict social media influencers, stand-up comedians and movie scenes.

Read: Meta launches standalone AI assistant

The darker consequence is that this will entrench Meta’s market domination. Exclusive knowledge of which ad works best across millions of campaigns gives it a unique advantage over rivals like Google, TikTok and traditional agencies. And while the obvious first users will be small businesses, there’s no reason why larger ones might not want to reduce their ad-agency spending for Meta’s tools if they’re cheap and effective.

Of course, measuring “effectiveness” means trusting Meta’s metrics, per Zuckerberg’s you-don’t-need-any-measurement just “the results that we spit out”. That is a little bit like trusting a casino that deals the cards and runs the tables to tell you how well you’re doing. It’s a convenient closed loop for Meta. With no independent verification, the company could optimise ads to boost user engagement and benefit its platform rather than for advertiser success, and brands would struggle to know if that’s the case or not.

Zuckerberg is meanwhile ushering in more bland and soulless content for his three billion Facebook users and two billion Instagram users. As more companies make ads that are “good enough” with his AI tools, their feeds will be flooded with algorithmic mediocrity and potentially more invasive ads. Years after Apple limited Meta’s ability to track users across apps for targeting, it can now exploit personal data to not just target users but manipulate them through an ad. As a mom, I’m forever seeing promotions for educational toys on social media. But, one day, an AI-generated ad might change itself if it sees I’m scrolling at 11pm and show a tired mother praising a product that gives her five minutes to herself. If my phone data shows I’ve been driving, a toy might be described as “perfect for car trips”.