Me and Mr. Tibbs

Thanks to the personal computer I’m failing to build using parts from my prior PC, I now not only have two nonworking computers, my small home office also has computer parts sprayed randomly across almost every surface. A photo of this workspace would be rejected by Slobs With No Self-Respect magazine, but could make the cover of Hoarders With a Problem. At least it has me thinking about what I would save and what I would trash if and when I ever do a legit cleanup of the shaken snow globe of trash that is my home office.

For example, there's that two-drawer filing cabinet that's been overstuffed and unused for [thinking ...thinking ...] well over 15 years. Not to mention the Le Brea Tar Pit of printouts surrounding my printer. I should just shovel them all into the recycle bin, but who knows if there's an article or an idea in there that I would be happy to find but would never miss if I tossed it unnoticed? The ability to memorialize ideas is one of the great benefits that print bestowed upon us and simultaneously burdened us with.

I'll stop this humiliating tour of my office because my actual point is that AI agents are about to turn trash into a resource ... and one that will make my office habitable again. Along the way, it will change what and how we know. In this regard, it will make us better at knowing, although in other regards, it could at long last get us to 1984.

For example, if I meticulously went through my filing cabinet, I’d find a folder for every talk I gave in the early years of the new millenium, including travel receipts and a printout of each slide deck, most of which were variations on a theme. If I read the contents of each folder, I would learn little that’d be helpful for my thinking today and lots that would make me feel even worse about myself than I already do: “I remember liking being in Hong Kong, but, wow, that speech rambled on!”

Calling Mr. Tibbs

Enter Mr. Tibbs, the personal AI agent I imagine having in a year or so. If Mr. Tibbs went through that filing cabinet, it would learn plenty. Of course, I’m imagining Mr. Tibbs version 4.0, which is not only smarter, but also magically has the physical mechanisms required to go through a stack of folders. In real life, or at least in real-ish life, I assume that United Parcel Service will be bought by Elon Musk, be renamed upX, and will offer a drop-in service that will scan your cabinets and incorporate all that information into your own Mr. Tibbs via the magic of retrieval-augmented generation and who knows what other technology by then.

What might Mr. Tibbs discover in those many folders? I don’t know, because I’m just a human. But I anticipate that it would find patterns in the shifting content of the slides, perhaps correlations with locations and sponsoring groups. Most of all, I hope it finds statistical correlations with what I’ve been writing more recently, enabling it to be helpful in what I’m currently working on because it “understands” the context of my path better than I do. With a more robust set of statistics about my prior work, Mr. Tibbs should have a fuller view of what I’m still not seeing and thus about paths I well might want to explore.