Yesterday I begin consulting on a project in the online advertising space. Today I am spending time getting up to speed on the domain and the client's business model. And I foresee darkness.
Here's the gist: the latest wave of online advertising philosophy is about empiric optimization, in which you create a huge number of variations on a campaign theme, try them out in emails, as banner ads, and adwords. Complicated algorithms analyze each ad's details: who clicked on them, how long they were on the page before they clicked on them, what search terms were associated with them when they clicked on them, and what site they were on and what other content was on that site when they clicked on them. This is all fed into a giant, impressive data thresher to evaluate what has worked, and it automatically adjusts placement strategies to favor those ads in those places that convert clicks to dollars efficiently and cheaply. Almost like evolution, the idea is to spread as many seeds with variations as possible so that the effective ones propagate.
While this still incorporates some manual creative efforts wrapped around a brand message, one can easily imagine that the next step is to replace creative effort with smart, genetic algorithms that generate ads based on a set of simple design heuristics. This image of a smiling, pretty woman against this backdrop of a blue sky with this 3D generated image of the product presented in this banner ad at this time on this site and look, we sold another 10 million in Peoria.
Here's the kicker. Once the ad generation, evaluation, and optimization is algorithmic, what you've got is a giant evolving, artificial intelligence that is optimizing itself to routing money from people to corporations. Artificial intelligence? Isn't that something that everyone's been dreaming of and working on for decades? Won't that be solved by computer professors and/or the military first? I am earnestly convinced that the advertising world will get there long before the academic world since they've got more money and time. It's also much more socially acceptable than military technologies, so there's nothing to stop its development right here amongst the populace.
This A.I. will evolve and get better and better at what it does. Corporations are all vested in ensuring that this happens.
This takes us, as so many things do, to speculative, dystopian science fiction. Before long we'll be in Asimov's world from "The Evitable Conflict" in I, Robot, i.e. positronic computers altering the surfaces of our world in ways too subtle to detect or resist in a perfectly-optimized pattern to get us to buy-buy-buy. And I'm pretty sure that no one in the ad world is concerned with that pesky first law of robotics. The A.I. won't be sure where to stop in its methods except in a parasitic don't-kill-the-host sense.
So my realization was to shift my expectation on how and where the war with the machines will be fought. When/if we finally have had enough and have to try to take this thing down, there won't be bullets flying and T2 robots stalking us on battlefields. It will be us scrambling with our eyes closed to scrub it clean from the world's skin.