The Jank Is Growing on Me
AI keeps getting more human. I'm increasingly into the part that doesn't
That cursed Zuckerberg selfie has been nagging at me again.
At the time, it looked like what happens when a $0.5tn company (now worth ~4x) spends a very serious amount of $$$ trying to render a human being & somehow lands on Wii Sports. These past few days I've been wondering whether the JANK had a point.
Charles Rollet
Muse is now topping the US app charts, and Meta this week followed it w/ Charm, a keychain device that gives the agent a face, a voice & basically somewhere to live. That doesn't mean << cute >> is why it took off. Free helps. Useful helps more. And Meta has never exactly struggled to get things in front of ppl.
The comeback arc is the obvious headline. What keeps pulling me back, tho, is the little cartoon & what it says about how we actually want AI to show up.
The more human these agents get in what they can do, the more value there may be in keeping what they look like obviously artificial. I'm hardly going to complain if an agent remembers things about me, books flights, moves money around, orders dinner & slowly works its way into parts of my life that used to involve another person. Exactly why I don't need the UI trying harder to resemble one.
A little character handles that tension pretty neatly. Give it a name, a voice, a face & a bit of memory and pretty quickly it feels familiar. But the cartooniness here gives me an escape hatch. I can get attached w/o having to pretend I've forgotten what it is.
We've actually been susceptible to this forever. Nobody thought a Tamagotchi was literally alive, yet plenty still took it far too personally when theirs died. And it doesn't take much. Even lightweight cues like a name or a voice have been enough to move trust in otherwise comparable systems. The framing alone can change how we read essentially the same machinery.
That matters more now because AI has ended up in a very weird lane.
We use it to write emails, search, code, plan trips, shop & increasingly act on our behalf. Then we hear about job loss, deepfakes, rogue agents and, if u scroll X long enough, the robots eventually killing us all.
At that point, a cheerful little blob doing the groceries makes a surprising amount of sense. It almost works in reverse: the more consequential the capability gets -> the more unserious the packaging can afford to be.
Which is also where I get a little wary of how well that works on me. The mascot can lower the emotional temperature w/o lowering the stakes. You're still handing over access, memory, money & judgment. Delegation just feels more casual when the thing asking looks like it belongs on a keychain.
Seen that way, the old Zuck image hits differently.
We spent years dunking on it because the avatar looked nowhere near human. Now that the software is getting much better at sounding like us, knowing us& acting for us, all that visible fakeness starts to feel oddly reassuring.
I doubt Zuck was playing that long a game. But maybe the jank was doing more work than we gave it credit for.
