Hardly a day goes by without a mention of Artificial Intelligence (AI), the Internet of Things (IoT), blockchain, or 5G these days.

Blockchain was more of a fad in 2017 than it is today, as many people equated the technology with soaring cryptocurrency prices representing a quick way to get rich for those who tried their luck. While at the beginning of last year cryptocurrencies went belly-up even faster than they had risen, the underlying technology's promises went much further than in the crypto space. Even though as a technology it is slow, energy-inefficient (even though this is gradually improving), and solves none of the problems that could not be solved by technologies that existed before it like centralised databases, digital signatures and certificates, fintech, escrow arrangements, letters of credit and information sharing on platforms like Git and Peer-to-Peer networks, blockchain enthusiasts still see a glimmer of hope in possible future technological improvements and favourable regulatory regimes.

In stark contrast, AI as it is today already provides a very stable foundation on which to build the future of things to come. Even if the technology had to abruptly stop being developed - a scenario which is unlikely to materialise given the huge investments that China is making in the field and the USA's pledges that it won't fall behind - the accumulation of data by itself would be enough to make existing platforms and algorithms better in an increasing number of sectors. Nevertheless, despite the great strides made at the level of technology, more advancements are still required before the technology is ready for deployment in several areas where it has the greatest disruptive potential. Consider, for instance, the case of autonomous cars. Computer vision (maybe computer 'sensing' is more apt as a term as this involves the simultaneous use of RADAR, LIDAR, SONAR, GPS and cameras), at this stage, is capable of a 99% accuracy rate. However, not many people will be willing to take the risk of being involved in a car accident every 100 times they get into their autonomous car and dictate a destination location.  

AI technology is advancing at break-neck speeds. Not only are new algorithms being developed to emulate human intelligence better, but new classes of microchips are being developed to be able to make AI computations much more energy- , process-, cost- and time- efficient as well. This, of course, begs the questions of whether our academic curricula are AI-proof, whether our labour force is flexible enough to withstand the labour market shocks that AI will surely bring about over the next decade, what will happen when AI is weaponised, and where Europe and Malta stand in all this, but these are questions that merit discussion at a national policy level and are best dealt with in those fora like those provided by the Malta.AI initiative and the various European Commission Workgroups on the subject.

Another pertinent question that this gives rise to is at what cost these new classes of AI-oriented microchips will be available. This is important because if the cost is low enough to make them easily replicable on end-devices themselves, most computation is likely to take place within the devices themselves, at the edge of the network that they form part of. If, on the other hand, they remain expensive, sensor-gathered information will need to be relayed for processing at very low latency to a centralised processing location, with the results then relayed back to the device itself. This is where 5G would come in handy as the standard promises very considerable bandwidth at very low latency (below 1 millisecond under ideal conditions). The 3GPP standard for standalone 5G has been around since June 2018, but consumer devices that are 5G-capable still need to be developed beyond the proof-of-concept sphere and dense, large scale deployments are still a few years away as part of the spectrum on which 5G will operate require a denser array of antennae than we have been used to even under LTE-A and fibre backhaul to be able to operate effectively.

Irrespectively of whether computations are handled by the IoT devices themselves or at a centralised location, the IoT, in combination with AI, promise a plethora of increasingly connected devices that will become smarter as time goes by. Cybersecurity, which is already a hot issue with existing IoT devices, is likely to become an even hotter topic as the stakes rise for hackers and end-users alike. To this end, the European Commission has just laid out some promising plans to beef up its cybersecurity agency ENISA that have been very long in coming, but which at least have finally managed to materialise.

Bernard Mallia is the CEO at Equinox Group and a European Commission external expert. The views presented in this article are Bernard Mallia's alone and do not in any way represent those of any of the entities for which he works.