The ubicomp manoeuvre was to realise that Moore’s Law cuts both ways.
Moore’s Law: pound for pound, computers double in power every 18 months. Or 100x over ten years. (And let’s forget about the death of Moore’s Law. We’ve been stuck in an AI overhang and I’m 100% sure there is a lab someone using AI to jiggle up silicon pathways literally as I write this.)
If computers get 100 times more powerful over a decade, we can EQUIVALENTLY say that computers get: 100 times smaller; or 100 times cheaper; or 100 times more abundant.
So that was the enabler behind Mark Weiner’s 1988 conception of ubiquitous computing.
At the turn of the century, a typical workshop or factory contained a single engine that drove dozens or hundreds of different machines through a system of shafts and pulleys. Cheap, small, efficient electric motors made it possible first to give each machine or tool its own source of motive force, then to put many motors into a single machine.
A glance through the shop manual of a typical automobile, for example, reveals twenty-two motors and twenty-five more solenoids. They start the engine, clean the windshield, lock and unlock the doors, and so on. …
Most of the computers that participate in embodied virtuality will be invisible in fact as well as in metaphor. Already computers in light switches, thermostats, stereos and ovens help to activate the world
Just the enabler though. The original conception, in Weiser’s Ubiquitous Computing #1, was inspired by looking at people and attempting to bend computing towards actual practice:
Inspired by the social scientists, philosophers, and anthropologists at PARC, we have been trying to take a radical look at what computing and networking ought to be like. We believe that people live through their practices and tacit knowledge so that the most powerful things are those that are effectively invisible in use. This is a challenge that affects all of computer science.
Ok.
BY ANALOGY:
If future AI models will be more and more intelligent (per watt, or per penny, or per cubit foot, whatever we choose measure) then we can equivalently say that, in the future, today’s AI models will become cheaper and more abundant.
What happens when intelligence is too cheap to meter?
Too cheap to meter: a commodity so inexpensive that it is cheaper and less bureaucratic to simply provide it for a flat fee or even free.
Me on fractional horsepower and fractional AI in 2012 and 2017:
Electrification began in cities around 1915 and with electrification so too came the potential market for washing machines, refrigerators, vacuum cleaners and a host of other commercial appliances. … By 1920, over 500,000 fractional horse-power motors were powering washers and other appliances in America.
In 2033, GPT-4 will be 100x faster, or 100x smaller, or 100x cheaper. What then?
LIKE:
Given an internal camera and access to Google, your oven cooks everything perfectly (and asks you about your preferences if it’s ambiguous).
Or telepathic light switches. Given some history and a bit of behavioural pattern matching, a Feynmann-level light switch could guess your intentions pretty well.
Viable home weaving of clothes? Buy infinite Uniqlo and a robot sewing machine. It’s interesting that ubiquitous AI means the end of ease of use: you can have a home fabrication unit or whatever that is as complex as you like, with whatever quantity of subprocesses, and there’s no skills gate that means you have to hire a trained professional or learn how to operate it. You will just talk to it.
Universal lie detectors (sub-visual readings of blood flow in the face, voice stress, etc).
In-home drones and robotics probably get solved pretty quick. So I should be able to say to a drone: “where’s that book I own that mentions X” or “where did I leave my keys” - but I’m not sure this is on the ubiquitous, fractional AI end of things.
Dunno, I need to think about this. Approach it systematically.
10 years after that, 100x smaller again. Nano AI, molecular scale intelligence. Drexler assemblers, smart dust. What if a grain of sand the size of the dot at the end of this sentence is as smart as you are, a handful for a penny, and can wave its flagellum to talk on the wi-fi.
The ubicomp manoeuvre was to realise that Moore’s Law cuts both ways.
Moore’s Law: pound for pound, computers double in power every 18 months. Or 100x over ten years. (And let’s forget about the death of Moore’s Law. We’ve been stuck in an AI overhang and I’m 100% sure there is a lab someone using AI to jiggle up silicon pathways literally as I write this.)
If computers get 100 times more powerful over a decade, we can EQUIVALENTLY say that computers get: 100 times smaller; or 100 times cheaper; or 100 times more abundant.
So that was the enabler behind Mark Weiner’s 1988 conception of ubiquitous computing.
Just the enabler though. The original conception, in Weiser’s Ubiquitous Computing #1, was inspired by looking at people and attempting to bend computing towards actual practice:
Ok.
BY ANALOGY:
If future AI models will be more and more intelligent (per watt, or per penny, or per cubit foot, whatever we choose measure) then we can equivalently say that, in the future, today’s AI models will become cheaper and more abundant.
What happens when intelligence is too cheap to meter?
Too cheap to meter:
Me on fractional horsepower and fractional AI in 2012 and 2017:
In 2033, GPT-4 will be 100x faster, or 100x smaller, or 100x cheaper. What then?
LIKE:
Given an internal camera and access to Google, your oven cooks everything perfectly (and asks you about your preferences if it’s ambiguous).
Or telepathic light switches. Given some history and a bit of behavioural pattern matching, a Feynmann-level light switch could guess your intentions pretty well.
Viable home weaving of clothes? Buy infinite Uniqlo and a robot sewing machine. It’s interesting that ubiquitous AI means the end of ease of use: you can have a home fabrication unit or whatever that is as complex as you like, with whatever quantity of subprocesses, and there’s no skills gate that means you have to hire a trained professional or learn how to operate it. You will just talk to it.
Universal lie detectors (sub-visual readings of blood flow in the face, voice stress, etc).
In-home drones and robotics probably get solved pretty quick. So I should be able to say to a drone: “where’s that book I own that mentions X” or “where did I leave my keys” - but I’m not sure this is on the ubiquitous, fractional AI end of things.
Dunno, I need to think about this. Approach it systematically.
10 years after that, 100x smaller again. Nano AI, molecular scale intelligence. Drexler assemblers, smart dust. What if a grain of sand the size of the dot at the end of this sentence is as smart as you are, a handful for a penny, and can wave its flagellum to talk on the wi-fi.