Op-Ed: Light-based chip brings AI to everything? Looking good

Posted Nov 20, 2020 by Paul Wallis
If you’ve been watching light based chip tech since it started, you’ve been watching for at least 20 years like I have. It’s been a very long haul. Now, in a sort of jackpot moment, light-based AI is doing everything, and doing it well.
Artificial intelligence Summit in Lisbon.
Artificial intelligence Summit in Lisbon.
RMIT Australia has been leading an international research program involving Australian, American, and Chinese researchers working on nanoscale imaging, machine learning and memory. What’s new about this is that it’s all on a single chip. This research is all about autonomous single platforms for multiple operations from drones to artificial retinas.
In addition to being a major achievement, the research is also likely to open the floodgates to a tide of new terminology, and this is only the start. This research includes words like “neurobotics” (yes, neural networks/robots, etc.) human/machine interactions, and scalable bionic systems.
Other possibilities include sensors in cars that can recognise objects on their own. This process is called AI decision-making. It can classify numbers, recognise patterns, and more. Better still, it can be integrated into existing systems.
Jump starting the future? Maybe.
What’s not so obvious about this situation is that it’s also a direct link to autonomous robotics and small and large-scale systems. The keyword here is “functionalities”. The functionality in the new nanoscale chip is based on black phosphorous which changes its properties in relation to different light frequencies.
Think about that a moment. Light comes in vast numbers of frequencies. There are 10,000 versions of the colour red, for example, and that’s just the visible spectrum. So, as this process is refined, the nanoscale reach of functions is likely to increase. That’s without any subsystems or layered systems.
That, it just so happens, is exactly what a multifunctional platform needs. Assign a value to a colour or a number. It’s that easy to do. That means an exceptionally wide range of possibilities for the nanoscale chip. It also means reducing the size of applications, a very functional hardware consideration.
This scalability is exactly what future tech always needs. From the days of computers which are now the equivalent of calculators taking up whole rooms or buildings, size has been an issue. Nanoscale high function chips are perfect for the sorts of futuristic intelligent tech which has been cluttering up science fiction for generations.
There’s an environmental benefit here, too. Nanoscale means fewer materials, lower energy requirements, and quite possibly even viable self-powering chips a la graphene. This is going to get really interesting, and from the look of it, soon.