AI vs. ads: surprisingly possible, surprisingly neglected
AI vs. ads: surprisingly possible, surprisingly neglected
Something interesting: people on the AI subreddit mostly agreed that AI poses a big risk to adtech - much more than I expected them to.
(I posted the poll as part of the AI vs. ads series.)
45% agreed that AI will destroy the adtech industry, 28% said no, 27% maybe.
I thought the results would be more towards no, given that 1) the wording of the question is aggressive and 2) the comments (weighted by upvotes) are mostly against the question.
My takeaway: surprisingly few people are analysing a scenario that surprisingly many people think is possible. And this scenario matters (a lot) because it would have a huge impact on how money flows in tech.
To put this into a framework:
It’s worth caring about a scenario if it has high expected impact, which is the impact the scenario would cause multiplied through by the chance of it happening. The (admittedly very weak) Reddit evidence is that the chance of AI destroying adtech is over 50%, counting half of the maybes, and the impact the scenario would cause is in the realm of $1tn of adtech revenue being destroyed or redirected. So the expected impact feels high, even if we lower both of those numbers a lot to be conservative.
But the cherry on top is that the scenario has a high level of neglectedness — in other words, it’s not being talked about much. Neglectedness makes it even more worth caring about a scenario, since few others do. Despite asking around in every forum I can think of and searching far and wide, I’ve come across very few analyses of how AI could be a huge threat to the existing adtech industry (if you’ve seen anything good written about AI vs. ads, please comment!).
Of course, a big disclaimer applies re: the selection bias and sample size of people on the AI subreddit (they are not superforecasters). To get something more substantial, I’m going to create some falsifiable predictions around the AI vs. ads scenario on Metaculus. However — I’m excited by the poll as a weak signal that this scenario is worth writing about.
Up next: a piece on the new Last Mile of content delivery that AI is (already) creating, and the risk it poses to adtech.