Superbowl
Last night it was the Superbowl, and the chatter this morning is all about two things: a guy called Bad Bunny, of whom I had never heard in my life until this morning, and the ads of Anthropic, with whom I'm much more familiar.
Here's the pick of the bunch from my perspective:
Let's start with a hat tip to the creative team. The execution of this ad is exceptional - everything from the casting, hair and make-up through to the sycophancy and just-noticeable latency with which we're all so familiar.
But like all great ads, it's not just the creative execution that lands so well, but the underlying message. It gets at the exact tensions I've been thinking about for a while when it comes to ad formats in AI Search results.
Put simply, the whole promise of AI is that it abstracts away the complexity of wading through 'ten blue links' to give the user a clean, synthesised, 'correct' answer. That's fine when it something objective where the prioritisation criteria are clear, or something non-commercial where the incentive to influence is less.
But for any consumer purchase journey, how can any ad-serving AI possible act both in the best interest of its commercial sponsors and in the best interest of the user? The pushback I often hear is that ads are useful if they are sufficiently personalised, but to be honest I think that's rubbish. The probability of the 'right' answer for the user and the 'profitable' answer for the AI company lining up perfectly feels very low to me. Sure, it's a utopia that may occur once in a while, but the more likely dynamic is that the well-resourced companies with scale will bully the longer tail for high intent real estate, just like the pay-to-play nature of supermarket shelf space.
The best practical response I've read on this came from Bryan Kim at a16z: Of course they're putting ads in AI.
Kim's central point argument is the correct one: that ads are a commercial imperative if we refuse to pay for things directly (as indeed is the case, evidenced by the <10% penetration rate of paying users on AI products).
That's is true but it doesn't resolve the question of whether Open AI remains a trustworthy arbiter of the truth for me as an individual in a post-ads world.
Kim briefly also tips into the "ads are great" thinking with which I disagree:
"The good news is that people actually do like ads! Ask the average Instagram user, and they’ll probably tell you that the ads they get are ridiculously useful: they get served products they actually want and need, and make purchases that actually make their lives better. Framing ads as exploitative or intrusive is regressive: maybe we feel that way about TV ads, but targeted ads are actually pretty great content most of the time."
I don't think that's true generally, and I don't think the Instagram comparison is the right comparator specifically. On Instagram, I'm mindlessly scrolling because I'm bored and seeking a dopamine hit, and seeing an ad for a product I might like tickles that fancy. In an AI search context, I'm more directed in what I'm looking for and want objective help in finding the best answer. In that context, the commercial incentives of Open AI putting their thumb on the scale is not useful to me.
Coming back to the Anthropic ads, they've done what all great advertising work does: create a conversation with the brand at the centre. The business model for AI is such a hot topic and it's a bold play to effectively rule out the business model that has underpinned the modern consumer internet (or at least risk massive shrieks of hypocrisy if you change your mind) but for now, in the context of such a heated AI race, it was a true "chef's kiss" moment.