Are robots with analog AI the future?
Digital brains are just too big, slow and unreliable to interact seamlessly with the fast-moving natural world. Human interaction may be the only phenomenon slow enough to make digital look impressive
Want to make a robot bee that can pollinate fruit?
Can you do this with a brain weighing only a few grams, with enough ‘physics processing horsepower’ to be able to locate, hover towards, land on and distribute pollen between countless flowers waving around capriciously in the breeze?
Nature, which masters this computationally formidable job effortlessly, doesn’t do it with ones and zeroes.
Preview, Natural Engineering Quarterly, Summer 2025
I know it was a long time ago, but who at the time could possibly have guessed that digital, which seemed like the answer to every technologist’s prayer, was nothing more than a stepping stone?
Just because we know differently now, we shouldn’t be so dismissive: we would never have been able to build today’s ubiquitous natural technology if we hadn’t been trying to perfect digital (and been so dissatisfied with its shortcomings).
After all, it was digital which gave us the tools to discover exactly how nature, in its relentless quest for efficiency, had perfected the means for creating mechanisms which knew enough about science to interact reliably and productively with the physical world in a way which eventually left digital technology looking more and more like a dead end.
For every pathway to efficiency, nature always seemed to be reminding us that digital was ‘the road not taken’ for good reasons: analog did things better.
Batteries included: but why?
Some will not be old enough to remember, but one of the reasons that our technology was so primitive in those bad old days, was because we had to struggle with such cumbersome things as ‘batteries’.
Nobody had yet stopped to fully explore how biology managed to do just about everything we needed to be able to do, but was able to do it without needing to use anything quite so ridiculously unsuited to the task of making intelligent things small, light and efficient enough to be sustainable.
PSA (which, if you come from outside the US means Public Service Advisory, an important warning about what follows).
- please don’t let what I am about to say here put you off watching the video
- the video above isn’t (just) about how to create a robot bee: he creates lobsters and eels too
The slides in this science talk video could quite possibly make you want to stop watching, even if you are perfectly able to cope with watching videos about technical stuff that you don’t fully understand.
The slide deck includes by far the most diverse range of extremely detailed multi-disciplinary technical content I have ever seen in any presentation (and I have worked my way through many thousands of science and technology talk videos in order to find presentations with innovation-related content worth writing about).
Don’t worry, fortunately neither the speaker nor the subject matter are anywhere near as offputting as the slides.
His theme (or at least one of many connected themes that he covers) is that nature provides clues about how to solve problems which are far beyond the reach of any present day computer and robot technology.
He uses biomimesis, but he doesn’t just use it just as designers tend to use it (i.e., mostly for inspiration regarding form and structure) he uses it for finding practical, functional scientific and technological solutions to some of today’s most intractable unsolved engineering problems.
The solutions that nature is offering us for the kinds of problems that it has already solved include replacing our present day robot’s digital computers with analog nanocircuitry and their electrical motors with synthetic muscles.
Insects have much smaller, lighter, lower powered, longer-lasting and sustainable onboard computers than anything we know how to build, and they interact perfectly with their surroundings, but their highly optimised biologically constructed computers aren’t digital.
Maybe digital is a dead end
We can now make tiny winged robots, but their brains, sensors and body motion control systems are extremely primitive in order to accommodate the bulk, weight and energy demands of digital computers and electromagnetically driven motion.
Nature has deftly sidestepped these constraints to create insect brains with extraordinarily sophisticated and accurate powers of navigation and capabilities for robust and flawless interaction with outstandingly demanding and rapidly changing physical conditions.
Joseph Ayers takes us on a roller-coaster ride through all the many scientific and technological hurdles that these challenges have presented, and how, by being prepared to learn how to implement nature’s tricks, he is building devices which have so many different innovations taken from so many different scientific disciplines that you can almost forgive what might just turn out to be the most daunting slide show ever presented.
Just one isolated example? I hereby nominate Ayers’ concept of ‘navigating by making scent visible’.
The talk is called: