1.
AI AND COVID: WHAT HAS BEEN, AND WHAT IS THE CONTRIBUTION OF AI, OF MACHINE LEARNING AND OF DATA ANALYTICS IN GENERAL IN THE FIGHT AGAINST THE PANDEMIC?
From science to support of governance policies, as never before, in this case, data has been at the centre of decisions.
The role Data Analytics has played has made a contribution at various levels. From visualisations which have made the scale and nature of the problem accessible, to the use of AI techniques.In terms of observation and forecasting as to how the pandemic has been evolving, superseding existing epidemic models (SIR and developments) in favour of what has been recorded "on the ground". In terms of the search for a cure, where bioinformatics has allowed us to cut normal experiment times, for example anticipating the structure of the virus. In terms of providing assistance in containing the virus, refining test kits for identifying the virus in individuals, or in monitoring the population vis-a-vis compliance with safety measures.
We have also had an active part to play in terms of some AI-based applications such as Safe Eye, which analyses video streaming to record compliance with social distancing and the use of face masks, or the implementation of a model to forecast how the pandemic is evolving in use at regional scientific committees.
From science to support of governance policies, as never before, in this case, data has been at the centre of decisions.
The role Data Analytics has played has made a contribution at various levels. From visualisations which have made the scale and nature of the problem accessible, to the use of AI techniques.In terms of observation and forecasting as to how the pandemic has been evolving, superseding existing epidemic models (SIR and developments) in favour of what has been recorded "on the ground". In terms of the search for a cure, where bioinformatics has allowed us to cut normal experiment times, for example anticipating the structure of the virus. In terms of providing assistance in containing the virus, refining test kits for identifying the virus in individuals, or in monitoring the population vis-a-vis compliance with safety measures.
We have also had an active part to play in terms of some AI-based applications such as Safe Eye, which analyses video streaming to record compliance with social distancing and the use of face masks, or the implementation of a model to forecast how the pandemic is evolving in use at regional scientific committees.
2.
WE ARE WITNESSING AN ACCELERATION IN DIGITAL TRANSFORMATION, WHICH IS NOW AT THE TOP OF ALL CIOS' AND CEOS' TO-DO LISTS. COMPANIES AND ORGANISATIONS IN EVERY SEGMENT SAY THEY ARE PREPARED TO USE AI AND MACHINE LEARNING TO REVOLUTIONISE THEIR PROCESSES AND TO IMPROVE THEIR BUSINESS. WHAT IS GOOD ABOUT THIS, AND WHERE ARE THE DANGERS IN TERMS OF THIS ACCELERATION?
Digital Transformation means rethinking processes and how you do business, but is only achieved when a company becomes “data driven”, learning from data, identifying what is of relevance in the data with a view to guiding the choices you make.
The capacity to learn lessons from large volumes of data which is varied in nature, the ability to successfully simulate behaviour inspired by human nature - these are the means for making complex easy-to-access and easy-to-use customised services accessible for everyone. Machine and deep learning, cognitive systems, conversational interfaces using natural language, prediction of phenomena and behaviours, comprehension of text, image and video or audio flows, affective computing, appropriately put together - these are and are becoming AI.
There is a lot, so much we are already able to do: recommendation systems, recognition of identity or conversation agents, to give some simple examples, improve the quality of our experience as clients, users or citizens. The risk, looking beyond the ethical implications and “bias”, is that the acceleration might end up being for the purpose of tactical manoeuvring, and, as is the case with all issues which represent a considerable interest, that it will not be supported with the right expertise, rendering ineffective tools and techniques which, conversely, can make it possible to raise the level of quality and the value of services offered.
Digital Transformation means rethinking processes and how you do business, but is only achieved when a company becomes “data driven”, learning from data, identifying what is of relevance in the data with a view to guiding the choices you make.
The capacity to learn lessons from large volumes of data which is varied in nature, the ability to successfully simulate behaviour inspired by human nature - these are the means for making complex easy-to-access and easy-to-use customised services accessible for everyone. Machine and deep learning, cognitive systems, conversational interfaces using natural language, prediction of phenomena and behaviours, comprehension of text, image and video or audio flows, affective computing, appropriately put together - these are and are becoming AI.
There is a lot, so much we are already able to do: recommendation systems, recognition of identity or conversation agents, to give some simple examples, improve the quality of our experience as clients, users or citizens. The risk, looking beyond the ethical implications and “bias”, is that the acceleration might end up being for the purpose of tactical manoeuvring, and, as is the case with all issues which represent a considerable interest, that it will not be supported with the right expertise, rendering ineffective tools and techniques which, conversely, can make it possible to raise the level of quality and the value of services offered.
3.
THERE ARE MAJOR THINGS EXPECTED IN 2021 VIS-A-VIS AI AND MACHINE LEARNING: TO WHAT EXTENT CAN THEY REALLY REVOLUTIONISE THE WAY BUSINESS IS DONE?
Let us try to give an example. There is a Utility that is aware of consumption by its domestic users, is able to predict their trends in the short and medium term, but knows little or nothing of the details as to how this consumption is generated. AI can single out the appliances which demand resources and open up options for customised services for the relevant user base. Anomalous variations in the way a specific electric device is being used can anticipate malfunctions with a view to offering maintenance services; the high rate of use of a electric domestic appliance could justify offering the customer to sell them, in billed instalments, a more efficient one; or else, you could automate delivery, for example, of detergent for a dishwasher that had completed 21 cycles since it was last replenished.
The electricity company is using his core business to open up new and lucrative product lines offering "extra commodity". The user benefits, improving their quality of life.
This is not the future we are talking about; this is what, in conjunction with and for the E&U market department at Engineering, we have made available with the Home EnergIA NILM (Not Intrusive Load Monitoring) service: a hundred machine learning models for singling out devices based on the signal sent to the meter. Is this not innovation?
Let us try to give an example. There is a Utility that is aware of consumption by its domestic users, is able to predict their trends in the short and medium term, but knows little or nothing of the details as to how this consumption is generated. AI can single out the appliances which demand resources and open up options for customised services for the relevant user base. Anomalous variations in the way a specific electric device is being used can anticipate malfunctions with a view to offering maintenance services; the high rate of use of a electric domestic appliance could justify offering the customer to sell them, in billed instalments, a more efficient one; or else, you could automate delivery, for example, of detergent for a dishwasher that had completed 21 cycles since it was last replenished.
The electricity company is using his core business to open up new and lucrative product lines offering "extra commodity". The user benefits, improving their quality of life.
This is not the future we are talking about; this is what, in conjunction with and for the E&U market department at Engineering, we have made available with the Home EnergIA NILM (Not Intrusive Load Monitoring) service: a hundred machine learning models for singling out devices based on the signal sent to the meter. Is this not innovation?
4.
HOW CAN ENGINEERING SUPPORT COMPANIES, AND TRANSFORM THIS ACCELERATION INTO A GENUINE PROCESS OF INNOVATION?
Engineering, with its Data & Analytics CoE, specialises in implementing highly customised solutions using a multidisciplinary approach and expertise, as Advanced Analytics and AI demand.
There is an inevitable process of democratisation with AI, but we should not be under any illusions. Obtaining disruptive results is something that depends on the ability of experts to operate following a methodology, to discover problems and to find the optimal solutions, proceeding iteration-by-iteration.
Thus, for example, one of our clients, Utility, is able, in their budgeting process, to rely on predictive models for takings and payments over a 90 day period, on a rolling basis, with a rate of accuracy approaching 100%. A bank uses our models for disposing clients toward cross- and up-selling in order to strengthen its marketing processes. A PA client transformed masses of documents into value, identifying and extracting the phenomena present, to allow them to be qualified, quantified and their scale measured. All performed automatically.
Engineering, with its Data & Analytics CoE, specialises in implementing highly customised solutions using a multidisciplinary approach and expertise, as Advanced Analytics and AI demand.
There is an inevitable process of democratisation with AI, but we should not be under any illusions. Obtaining disruptive results is something that depends on the ability of experts to operate following a methodology, to discover problems and to find the optimal solutions, proceeding iteration-by-iteration.
Thus, for example, one of our clients, Utility, is able, in their budgeting process, to rely on predictive models for takings and payments over a 90 day period, on a rolling basis, with a rate of accuracy approaching 100%. A bank uses our models for disposing clients toward cross- and up-selling in order to strengthen its marketing processes. A PA client transformed masses of documents into value, identifying and extracting the phenomena present, to allow them to be qualified, quantified and their scale measured. All performed automatically.
5.
JUMPING AHEAD A BIT INTO THE FUTURE, IN 2025 TO WHAT EXTENT WILL AI HAVE ENTERED INTO OUR LIVES?
In recent years predictions have systematically been exceeeded by reality. AI is becoming pervasive: we interact with customer care via conversation agents; image analysis systems single out potentially risky items at a station or a bank; digital recruiters evaluate a job candidate's characteristics and provide a profile of their suitability.
To an ever-increasing extent, our lives will be impacted by AI-generated decisions (granting a loan, the risk associated with a user, a tax inspection decision etc.), in which transparency of the relevant decision criteria will become ever more fundamental, so they are not discriminatory, or downright harmful. Topics such as Explainable AI or Responsible AI are becoming ever more relevant as these technologies increasingly spread.
In this respect, Engineering, thanks to its approach to AI, focusing in-depth on models, is certainly at an advantage.
In recent years predictions have systematically been exceeeded by reality. AI is becoming pervasive: we interact with customer care via conversation agents; image analysis systems single out potentially risky items at a station or a bank; digital recruiters evaluate a job candidate's characteristics and provide a profile of their suitability.
To an ever-increasing extent, our lives will be impacted by AI-generated decisions (granting a loan, the risk associated with a user, a tax inspection decision etc.), in which transparency of the relevant decision criteria will become ever more fundamental, so they are not discriminatory, or downright harmful. Topics such as Explainable AI or Responsible AI are becoming ever more relevant as these technologies increasingly spread.
In this respect, Engineering, thanks to its approach to AI, focusing in-depth on models, is certainly at an advantage.
Marco Penovich
He has always been involved in analytic systems, curious about everything concerning this world. He's interested, with an engineering approach, in the method, design and modeling aspects.
Marco is the Head of Engineering's Data & Analytics Center of Excellence, a group of over 150 professionals with the aim of giving substance to the term "data driven". Data Management, Advanced Analytics, Data Visualization and Data Governance are the themes on which the skills of a multidisciplinary team that operates across all markets are based
Marco is the Head of Engineering's Data & Analytics Center of Excellence, a group of over 150 professionals with the aim of giving substance to the term "data driven". Data Management, Advanced Analytics, Data Visualization and Data Governance are the themes on which the skills of a multidisciplinary team that operates across all markets are based
Contact us
marketing@eng.it
Tel. (+39) 06-87591