Sharing Ideas

Will artificial intelligence train humans?

I’m not a teacher.

However, training has always been an important part of my professional and private life. I remember my first job as a consultant in the field of food safety. This was at the beginning of HACCP1, and part of my job involved training confectioners, canteen, restaurant and bar managers on the methods used to prevent food contamination. Twenty years later, I now deal with research and training in the HVAC/R sector. During my career I have taken part in countless activities as a “student”, due to the close relationships between my company and various universities and participation over the years in numerous international meetings around the world focused on technology and knowledge sharing.

My wife works as a high school chemistry teacher. My daughter, maybe because of this, or due to her age or personality, has always felt the need to have a tutor figure even outside of school: gymnastics coaches, music teachers, skating instructors, English teachers. The relationship between trainer and trainee has always been part of her role-playing games, either alone or with friends.

Can I therefore conclude that training plays a crucial role in my life and the life of others?

1This stands for Hazard Analysis and Critical Control Points, and entails a set of procedures aimed at ensuring the safety of food.


The key role of training and competence in the HVAC/R sector

This is a rhetorical question. History shows that training or education, the passing on of knowledge between generations, is one of the cornerstones in the survival of humans on this planet.

In my speeches on this topic, I often like to mention some recent “evolutions” focused on innovation and training. For example, Amazon.

Amazon was founded in 1994 as an online bookseller. Thanks to the vision of its founder Jeff Bezos, “being Earth’s most customer-centric company”, Amazon has evolved over time to become the giant that everyone knows. Amazon Web Services, Logistics, Echo and Prime are the steps in its success, including the purchase of a supermarket chain in the United States in 2017. From books to web storage services, shipments, home automation interface devices, content streaming and food retail. Every step has been accompanied by a massive staff training and education campaign, including technologies, business models, mindsets and the soft skills needed to address markets that were unknown to those who started out delivering books.

In the HVAC/R sector, I was very impressed by a study dating back to 2010 by a non-profit association called Carbon Trust, funded among others by the Scottish government. The study, called Refrigeration Roadmap, analysed the scenario for commercial refrigeration worldwide in 2010-2022 from the point of view of environmental sustainability.

It is interesting to read again now how back in 2010 the key technologies highlighted for the future were those that are now consolidated. These included LEDs, natural refrigerants, water loop systems, inverters, evaporative cooling, floating load management and the use of doors on refrigerated showcases.

Magnetic refrigeration, internet shopping as an alternative or addition to food retail (e.g. Amazon GO), types of alternative food and "cold room" supermarkets where consumers pick-up their orders without entering the store were also considered. In other words, a complete examination that envisioned elements that still seem futuristic and have yet to be fully investigated.

In all of this, one important aspect was represented by a bubble chart. One large red disk underlined how "Training" was considered one of the tools for environmental sustainability that was really worthwhile investing in immediately.

source: Refrigeration Roadmap, 2010, Carbon Trust

It is not hard to find evidence to support this claim. All the more so in a situation like the one in 2018, where many, for example see here, have stressed the lack of competent technicians in the HVAC/R sector and the critical nature of the generational transition we are going through now.

Too bad that there was no explicit mention of one of the current technological trends, IoT.


Digitalisation and humanisation

On this blog we have already talked about machine learning. The ability of a device, essentially made up of hardware and software, to independently learn and then make available what it has learned to offer a service to users. In HVAC/R, we are observing and developing the concept of digitalisation with the aim of connecting together air-conditioning and refrigeration units in order to provide the data needed for learning to machine learning systems. The services offered range from preventive maintenance to system optimisation.

In this stage, however, such artificial intelligence needs a human tutor. I’m not talking about programmers, but rather people who, being experts in applications, can "pass on" their own experience to these artificial systems. In fact, machine learning creates correlations and assumptions based on an enormous amount of data (so-called big data), which represents its "wealth of experience". The resulting significant impact in terms of time and storage space is the main obstacle to its widespread use. Human tutorship, then, can achieve results with less effort. An expert in HVAC/R techniques would be able to suggest the probable existence of correlations, for example, between the data from a vibration sensor (accelerometer), an oil or refrigerant level sensor, operating pressures and other measurements, and correct or incorrect operation of an air-conditioning unit. Software programming can then be used to transform human experience into logical guidelines for information systems.

In this way, refrigerant leaks, the freezing of evaporator coils, the fouling of condensers, the accumulation of oil in pipes and many of the most common problems encountered in HVAC/R can be learned in less time by computer systems, so as to provide a more precise detection service based on human teaching.

So when will we be ready for the opposite process? When will it be possible for artificial intelligence, to train, to pass on experience or to teach a human being?

This topic is already being discussed. Today, insiders seem skeptical, as teaching is a purely human activity and artificial intelligence is considered at most as probable assistance.

However, I feel that I have experienced a form of learning in which machine-learning based systems can provide effective tutorship to humans. The services of IoT Analytics (Descriptive, Predictive and Prescriptive) provide a preliminary vision of a scenario that will most likely occur, considering the background conditions of the recent past. These allow, for example, a refrigeration system maintenance technician to deal with problems in the field in advance. They remind me of the teaching techniques of parents who tell their children, whose memory is not yet developed, of the events of the past, show the current scenario and offer a glimpse of the future. It is up to the children to use this information actively (learn), passively (obey) or superficially (ignore).

I realise that this kind of artificial systems are in no way "humanised" and therefore not very appealing as teachers. However if alongside future autonomous HVAC/R systems with self-diagnostics and self-repair we will find expert technicians who have learned from artificial systems, I will feel more confident in the possibility of a business that is finally environmentally sustainable.


No comments yet. Be the first.