Can AI power nudgeware?
Despite all the enshittification and walled gardens of Big Tech, I’m an optimist. In 2016, I made the case for something I called nudgeware. I defined this as:
Software that changes our environment so we are empowered to make better decisions, without limiting our choice in any way
Things like ad blockers, website blockers and time-trackers were nudgeware-ish, but they seemed like blunt instruments to me. I felt something was missing: a layer of human-computer interaction which sits between us and all other software, helping us act more in line with our intentions1. I still don’t believe a compelling example of that layer exists (even though I tried to create one).
But AI may be the technology that causes a nudgeware breakthrough. AI can:
Interpret human intentions
Align with them
Remember them
Be an agent for them
Control other software
Perform complex operations
Put these to the service of nudgeware and you could get a layer of software that:
Helps you do things you intend
Helps you not do things you don’t intend
How might it work? Like this. How might it look? Like Jason Yuan’s MercuryOS.
But there are big challenges:
OSes are expensive to make and distribute
Existing OSes and software limit AI controls (technically or legally)
Greater controls for AI = greater risks
AI costs money to run
I’ll address these challenges in future AI vs. ads articles.
If I came up with this concept today, I’d probably decrease the focus on ‘environment’ and put intention at the heart. So a nice 2024 update to nudgeware might be intware or intentware. Intware has the advantage of connoting a few other related ‘ints’, like interoperability, interaction, interpret. Although that’s not so convincing when you consider that 1k+ words start with ‘int’, including ‘interphalangeal’ and ‘interparoxysmal’.