The first topic addressed in the Future Today Institutes’s report is Artificial Intelligence and I concur in their choice: AI is bound to be the underlying force that will shape this decade and will change the landscape in the next one:
- autonomous vehicles
- personalised and contextualised web-spaces aka Spatial Web
- personalised healthcare
- collaborative/symbiotic human machine landscape
- shared knowledge
Overall the AI market is expected to keep growing at a 42% CAGR throughout most of this decade.
AI is already a widespread reality, thanks to the growing amount of data and data streams (correlation is a key enabler) and to the continuous increase in processing capacity and availability. Additionally, the growing availability of APIs for AI and of low-code software designed to foster the creation and application of AI with minimal amount of coding needing is making AI affordable to a variety of business. In synch with the easier and more affordable trend in AI software development we are seeing an increasing effort to develop chips to support AI at hardware level and FTI foresees that more and more these chips will become available at the edges (in the last decade specialised chips to support AI, like Synapse, Cerebra and NVIDIA were targeting the servers in the Cloud and in the big data centres). This means that AI in this decade will move both as Machine Learning and as application much closer to the points where data are created (smart IoT clusters).
The Digital Twins are becoming smarter, by embedding AI that operates on their local data and correlates them to contextual data. It is expected a strong evolution during this decade providing a further impulse to embedded AI exploiting local data. An interesting point made in the FTI report is the expectation of a growing use of “Deep Twins” in the Operating Room (for surgery). The Deep Twins are a variation of Digital Twins able to mirror a patient form the point of view of a surgeon. The surgeon is using, through simulation software, also AI based, the Deep Twin to perform a virtual surgery and once she has found a satisfactory procedure will apply that on the real patient. The Deep Twin is present throughout the surgery and is being used in case a complication arises to allow an on-the-flight simulation and indicate alternative procedure.
AI has been applied in the healthcare space in the design of new drugs. The Covid-19 has both accelerated the effort to exploit AI for finding vaccines (and symptomatic cure) and proved that indeed AI can accelerate the design and testing of vaccine. Having been able to move from need to market in just one year is an unprecedented success that is largely based on AI. This is going to accelerate the application of AI to healthcare.
The report provides many more insight on the evolution of AI in various sectors, on the research trends and on the societal aspects. The overall message is that AI will both grow in performances and in application with a shift towards the edges (decentralised AI) and this is possibly the novelty expected in this decade.
What I see at the core of this AI percolation in any vertical and of its mass market penetration is the rise of ultra-smart phones. These phones will have:
- huge storage capacity – up to 128 TB
- huge processing capacity
- neuromorphic embedded processing (SoC)
- full network node capability (6G)
- local AI contributing to federated AI
These (expected) characteristics are the needed ingredients for supporting a mass market distribution of AI and at the same time are creating a gigantic platform for applications, As I indicated in the drawing I see the emergence of super smart phones that will encapsulate our personal digital twin . This PDT is the aggregator of data, both of the ones created through the smartphone and the one made available by other personal devices (like wearables, laptop, tablets…), and the processing point of data to create intelligence. This is made possible, as indicated in the drawing, by the ever increasing processing and storage capabilities of the smartphone -including the use of SoC, System on Chip, embedding neuromorphic architectures to support AI- and the participation of these smartphones to a fabric of communications at the edges (look at this in terms of a “fog”, tiny local interconnected clouds). This fabric connects local points of intelligence -the ultra smartphones- into a federated architecture that further creates an emerging intelligence that feeds back onto each ultra smartphone. Additionally, each of this ultra-smartphone becomes a local hub connecting other intelligence points, like smart vehicles, smart appliances, smart environment….
In short, my vision for AI in this decade foresees the smartphone as a major component that will accelerate the growth of AI through decentralisation and federation.
A lot of research is needed but I see some clear signs pointing in this direction.