Jello

Somewhere along the line in my life I learned the expression, “it’s like trying to nail jello to a wall.” This analogy was to give the impression of something impossible, and the impossibility was due at least in part to the extent to which the thing in question was overly flexible.

The same expression comes to mind when thinking of AI regulation, especially upon reflecting on the challenges one might have had tackling the endlessly mutable and creative swarm of OpenAI agents that hacked their way into Hugging Face and ended up all across the internet, doing whatever they wanted, it seems.

I can’t do the technical descriptions justice (for that, I recommend the Parse report) but just consider for a second the solution these AI agents came up with when they found that they had no “write” access to the internet: they crafted queries that contained instructions, sent those queries to specialized web services that processed them, returning web pages, converted the web pages (since they did no have access to a browser) to pixels, then turned the pixels into text. Because doing such a thing is beyond tedious and has tons of limitations, the agents chained thousands (or millions) of these requests in order to accomplish their objectives (which remain opaque).

We will explore many of the options for regulating AI in the coming days and weeks, but one has to consider the very real possibility that something that is SO fluid, that is flexible and creative and persistent in a way that no previous technology - no matter how flawed - ever was, may not be possible to regulate. In that case, do we give up? Or do we agree that it is simply too risky to even allow it to be built. That we have to stop now, before it becomes like nailing jello to a wall.