Disconnection and pausing AI

As the world continues to digest the implications for our future of AI, one thing seems clear: to the extent that AI runs on data and connectivity, the best way to deprive it of power and to stay its path towards potential misadventure is to deny data, first and foremost by disconnecting. This flies in the face of the general trend towards ever greater and denser interconnectivity through the datafication of everything and the Internet of Things (IOT).

We can do this by returning more and more to legacy modes of content ingestion: print media, live performances, going to movies, connecting with people one on one, etc. Many of these are things people intuitively love and miss. Of course, to do these things at scale is carbon-intensive. But particularly during all-important political campaigns, we need to do more of these kind of things to minimize the risk of deep fakes permeating by bots across the decentralized distribution networks provided by social networks. Pushed to an extreme, it's not too hard to imagine generative AIs customizing content for each of us optimized to drive us bonkers: imagine an AI that knows exactly what your buttons are and can generate a Donald Trump giving speeches designed to bug you the fuck out. No longer the stuff of science fiction.

Take for instance cars. I've long rued the fact that dependence on Google Maps (often just to optimize route for traffic conditions at a given point in time) has cut into my ability to get around places unaided. This deskilling will get worse with self-driving cars. As with everything else, the trend with cars is towards greater and greater interconnection. Electric cars have been dubbed "supercomputers on wheels." Teslas already update their software and deliver fixes over the cell network (as they did a few years back when there was a hurricane on the East Coast and Tesla updated its cars' firmware to unlock extra battery capacity that was reserved for more expensive models).  If they are linked to a network by 5G, think about the possibilities for mayhem if a rogue AI goes HAL on us. I hope Musk is pondering this.

As it is, we need to be ever more careful about giving our data away for free. (says the guy who just posted on a Google platform)