Until now only 2% of the world's economy is affected by AI, assumes Jürgen Schmidhuber, the head of the Swiss Dalle Molle Institute for Artificial Intelligence.
Today, AI is mostly used in marketing, Schmidhuber states. Amazon does more than one third of its revenues due to AI predictions - people who have bought this bought also that (Sterne). “The next wave of AI will be bigger” and will comprise production, health industry and more, Schmidhuber predicts.
Like Indiana Jones' search for the holy grail, many in the MEMS community, scientists, start-ups and industry are looking for the “killer application” for MEMS devices. The coming killer app was discussed regularly and extensively on many Commerzialization of Microsystems Conferences such as COMS conference and elsewhere.
After introduction in military applications, the automotive industry historically paved the way for massive use of MEMS applications in assistance systems for safety features and now your smartphone with lots of MEMS devices. Will AI bring the next killer applications with a little help of MEMS?
AI is not a brand-new topic, it has been discussed since the 1940s. But today it is the hot thing even in many business events with IVAM Microtechnology Network or at University of Applied Sciences Hamm-Lippstadt.
The big step forward for AI came with larger computing capabilities. With quantum computers in the nearer or farer future those capabilities will be raised to even higher power.
Before we discuss whether AI will be part of the next MEMS killer app, it might be interesting to look at the main forces driving business in the era of digital transformation.
If we leave behind the sometimes confusing definitions about digitization, digitalization and digital transformation and talk about the fourth industrial revolution or "industry 4.0" the consulting company Roland Berger (Roland Berger / BDI 01.02) has identified four major forces in this transformation process.
These forces are part of the AI game and its implications for the MEMS industry and we have to keep them in mind for the coming. To quote Richard Buckminster Fuller: “Don’t fight forces, use them”.
“The true sign of intelligence is not knowledge but imagination” and “intelligence is the ability to adapt to change”. You will find many other definitions about the nature of intelligence on the net. The quotations cited are no general definition.
Intelligence can only be measured indirectly with a test and be concluded from there. Intelligence is what an intelligence test can measure. Human intelligence is not constant. When you do an intelligence test tired or drunk, it will result in a lower IQ.
There are also different “disciplines”:
When it comes to AI Jim Sterne says: “Artificial Intelligence is a program that can reprogram itself”. Programs can learn to detect patterns, process language, can convert text to speech (e.g. Amazon Polly) or speech to text, translates texts (deepl.com), plans optimized travel routes for delivery, automizes text generation for products specs (retresco.de).
In Xian, China, I was communicating live with many chinese people speaking no English and me knowing only “Nihao”. We exchanged our views with the help of AI powered translation assistants.
If you describe the coming applications of AI as a rocket, the fuel is data. So you need lots of data to get somewhere and it is not easy to steer the rocket (see force one above).
AI often fails with tasks a four-year-old easily can perform, for instance understanding a joke or irony. AI also has problems with subliminal messages: If an English person tells a German person "it might be good idea to make an appointment with me some time", it could be, that the English person is really upset and wants to talk to you immediately. A German person and an AI-based program might misunderstand this intention.
Ask the Apple based Siri: “Say something dirty” it will come up with “mud”, “compost”, “pumic stone” and “sludge”. Try it out!
Many companies are promoting sensors “with a degree” for a long time. Smart sensors, that not just collect data but also do something intelligent with it. Recently, the German company Bosch presented the video #LikeABosch at CES in 2019.
Internet of Things (IOT) is now mainstream and the massive interconnection of devices and the enormous collection of data is a precondition of AI applications.
A major trend now is the screenless internet with voice assistance such as Amazon Echo, Google and other services. Numbers of sold devices are skyrocketing. Amazon had sold 100 Million devices until now and there are more to come.
Today when little kids sweep with their fingers on the front of big fish tanks to alter the view and are puzzled because nothing changes, kids in the near feature will grow up talking to machines and wonder why some stupid thing like fish tank doesn’t answer.
I learned a new term that describes the transfer of expectations towards a product or a service from B2C to B2B business: It is called „Amazonification“ If you are used to talking to Amazon echo, you might expect similar things from your next car or your machine in the workshop. For example modern turning and milling machines talk to you, but they are not interactive. That will come with a little help of AI.
Talking to machines might be the next trend for industrial machines. MEMS-microphones can filter out interfering noise by noise cancellation and therefore they support many applications in loud environments.
So AI is not the next killer app for MEMS. As the search for the holy grail was somehow senseless, as shown in the Monty Phytons “Knight of the Coconut”, there is no MEMS killer app.
But I would bet on MEMS-Microphones in any machine or device for screenless internet-applications!
Berger, R. ; BDI (01.02): DIE DIGITALE TRANSFORMATION DER INDUSTRIE 2015