“What a computer is to me is it’s the most remarkable tool that we’ve ever come up with, and it’s the equivalent of a bicycle for our minds.” ~Steve Jobs
One of Steve Jobs’ favorite analogies was between a bicycle and its intellectual equivalent – the personal computer. Jobs brought this up often over the decades, and his point was that the personal computer is a tool that gets us more intellectual distance per thought cycle. Apple ran several elegant ads over the years alluding to this parallel.
Jobs drew inspiration for this concept from a 1973 article in Scientific American about the origin and impact of the bicycle. The article shows a chart of the energy cost of locomotion for various animals, derived from research by Vance A. Tucker at Duke University (“Energetic Cost of Locomotion in Animals”). Tucker’s research suggests that humans are not the most efficient animals on a “Calorie Per Gram Per Kilometer” basis (a horse, for example, performs more efficiently). However, when the bicycle is introduced, humans are far more efficient than any other animal.
The parallels between bikes and personal computers run deep, not just with regard to the benefits of computing, but also with regard to the drawbacks. One thing Tucker’s study did not take into account was the amount of calories expended to build the pathway that the bike travels upon. While a horse can travel through a forest without expending the calories of paving a road, a traditional road bike would struggle. The computing equivalent to a paved road is digitized information. In order for information to be relevant to a computer, it needs to be translated into bits. Only after the subject material has been digitized can the computational benefits be felt.
Further, while a human on a bike can quickly move linearly, a bike struggles with lateral movement (side step) and vertical movement (it cannot easily jump). Computers, similarly, are more native to mathematical paradigms of “thinking” rather than more creative paradigms. Computers are, fundamentally, computing machines and their foundations are in mathematical logic. Abstractions of this basic mathematical logic include addition, subtraction, multiplication and division. It is not a coincidence that the first uses of computers were for mathematical computations. Only after software introduced more layers of abstraction could a computer traverse more complex terrain (much like the gears, shock absorbers, and other innovations in bicycles).
In a previous article, I stated that artificial intelligence is not magic. Rather, it is the natural progression of the computation revolution in a similar way that the internet has been a natural progression of the communication revolution. The “bicycle for the mind” analogy fits well with the computer’s descendent – artificial intelligence.
To trace the progression, let’s consider a single cognitive task – transcription. In the era before mass personal computation, the majority of people were limited to handwriting. Personal computers, as the bicycle of our minds, introduced typing to the masses. Just like a bicycle, though, learning to type took some getting used to. In the era of AI computation, we will have extremely high-quality voice transcription. Searching for conclusive studies on the speed of each of these three paradigms yielded no comprehensive studies. Ad hoc research suggested somewhere around 20+ words per minute (“wpm”) in normal-quality (neither rushed nor particularly neat) handwriting, approximately 40+ wpm for typing and 100+ wpm for speaking. As a sanity check, I tested myself and my numbers were roughly in line (50wpm- though I have very messy handwriting, 75wpm and 175wpm).
Like a bicycle, personal computers roughly double the cognitive task of converting thoughts into written words. And, like an e-bike, AI computation roughly doubles that speed and productivity. Hundreds of cognitive tasks that have benefited from the personal computer revolution are poised to further accelerate – because of AI.
AI is still early on. One of the key differences between a bike and an e-bike is the motor and electrical needs. An e-bike is more complicated and requires more maintenance (just ask the leadership at Bird). AI requires “training” the algorithms (perhaps analogous to the software and hardware that calibrate an electrical motor) and an infrastructure to charge the e-bikes. AI also unlocks new potential and so necessitates concepts like algorithmic regulation (analogous to governing the max speed on an e-bike). Just as primitive computers initially were stuck in the realm of raw calculations, so is AI largely still in that realm. However, continued progress will mean that in the years and decades to come we will all benefit from extra power as we ride our intellectual bicycles.