Issue link: https://maltatoday.uberflip.com/i/1236030
16.04.2020 6 INTERVIEW Prof Alexiei Dingli is a Professor of AI at the University of Malta and has been conducting research and working in the field of AI for more than two decades, assisting different companies to implement AI solutions. He forms part of the Malta.AI task-force, set up by the Maltese government, aimed at making Malta one of the top AI countries in the world What impact will AI have on the world of work over the next 10-15 years? e world of today is rapidly changing, and we expect to see an incredible change in the coming decade, thanks to AI. Let's not forget that one of the positive effects of the pandemic is that people are be- coming more reliant on technology, are learning to live online and are no longer shy of technology. is circumstance in itself leads more people to use and ulti- mately trust technology. Employers too are undergoing such a mental change. For the first time in their lives, their business is grinding to a halt! In the coming years, AI will not neces- sarily replace entire jobs, but it will take over specific tasks. It has already started replacing these tasks, and this will keep on accelerating in the coming years. Ac- cording to the World Economic Forum, in 5 years, automation will account for more than 50% of the tasks in existence. We believe there are four potential scenarios which we should consider: • Some jobs like the caring profession (nurses, doctors and others), the cre- ative industry (actors, musicians and others) will not be affected much by AI because AI is still not very good at these tasks. • Other jobs will become obsolete. ese include drivers (because of the rise of self-driving cars), factory work- ers (because of automation) and many others. • Many new jobs will emerge. ese are the jobs which we haven't even invent- ed. It could include services like Organ Creator whereby people design new organs (such as a new heart, kidney or liver) which would replace defective ones. • Most of the existent jobs will change since AI technologies will augment their fundamental tasks. e effect of this will be massive. It might be as small as helping stackers in a super- market identify expired products us- ing augmented reality, up to assisting a surgeon during an operation. What will or won't AI do? AI is extremely good at handling re- petitive tasks at an incredible speed and with amazing precision. Because of this, it can easily replace almost any task which has these characteristics. Just think of a self-driving car, the driver has to look at what's happening around him (repetitive job), he has to notice any anomalies im- mediately (at incredible speed) and take decisive action (with amazing precision). By looking at these characteristics, one can quickly identify whether a particular job is at risk or not. On the other hand, AI is still not good at handling people. When people communi- cate, they do so not only through speech but also through social norms. ings like turn-taking, gesturing, showing empathy and many other unwritten rules. ese are things which we learn as we grow within a community. But for a machine, these things are not natural since it lacks social context, and someone has to teach them to it. Human-Computer interaction made giant leaps in the past decade, but we're still far from achieving the natural- ness of human communication. One can't expect to get a robot to take care of a hu- man patient because it will not be able to handle the situation, not the medical side of course but rather the human side of it. Why should specialists be worried about AI? Specialisation has always been the do- main of humans because to achieve it; one needed years of studying combined with experience. However, the tables are now turning. e internet contains much information, thus allowing AI systems to process that information and learn from it. is process is equivalent to having a human studying and gaining experience over time. e only difference is that for a machine, this happens within a few hours or days. If we just look at one example, in 2018, a group of 20 lawyers specialis- ing in corporate law were pitted against an AI. ey were each provided with five non-disclosure agreements (NDAs), and they had to identify any faults. e law- yers, on average, took one and a half hours to complete the task and achieved 85% ac- curacy. e legal AI, on the other hand, obtained 94% accuracy in 26 seconds. e difference is massive, and it shows that every job, even high paid jobs, will be affected by AI. Of course, we will not be seeing robot lawyers debating in the courtroom any time soon, but AI systems will start assisting lawyers in their regular tasks. What can managers do to keep up with the times? Malcolm X once said that "the future be- longs to those who prepare for it today". Some people are already dreaming and engineering the solutions of tomorrow. e technology is ripe, we have the pro- cessing power, and there is finally the po- litical will to embrace the AI revolution. It is only a matter of time before it takes over. So Managers need to take the first step and learn about AI. If in doubt, they should seek an AI expert, and they'll prob- ably be surprised at the enormous savings which AI can bring to their organisation. e process in itself is rather straightfor- ward as well, and it follows some well-de- fined steps. First of all, it is important to start educating people about what is AI. e education process has to start from the very top, down to all the levels of the organisational hierarchy. Second, the or- ganisation undergoes an AI readiness audit to ascertain the state of the organi- sation with regards to adopting AI. en, a consultation process is carried out with all the stakeholders to identify potential AI case studies. ese case studies are then ordered together with management based upon feasibility, the investment required and potential returns. en management sim- ply decides which projects to implement. Whether they are implemented in-house or outsourced, really depends on the complexity and the expertise available in the organisation. AI experts will be there to oversee the running of the project. On completion, another more ambitious project is select- ed and executed. e cycle continues like that until the organisation manages to augment its processes with AI. How will the future of work look? Work as we know it will gradually change. Reliance on automation will in- crease exponentially, and we will forget how things happened in the past. Just think of a smartphone; it replaces a lan- dline, answering machine, a camera, an alarm clock, a satellite navigation system, a scanner, a music player, a calculator, a TV and loads of other stuff. Young people have probably never seen some of these old devices unless they visited an antique museum. e same will happen with AI. Our soft- ware will get an injection of AI features capable of processing vast amounts of data. e computer will become omni- present; no longer housed in a small box but accessible from anywhere within a lo- cation. e interface will be easier for an- yone, allowing seamless human-like voice interaction. So work will shift from the harvesting, processing and presentation of information towards the management of AI services which will take over most of these tasks on our behalf. With sophis- ticated robots on their way, even manual tasks will switch towards semi-manual tasks. Automation will also make a lot of manual processes redundant. Where this is not possible, exoskeletons will help hu- mans carry heavy loads without running the risk of jeopardising their health. What is the most impressive artificial intelligence we have today, in your opinion? e most impressive AI today is with- out doubt computer vision. It is the single subfield of AI where the computer man- ages to beat human levels consistently. Be it the identification of cancerous cells in medical images, the recognition of objects or people in a video, or just in- terpreting the road for self-driving cars, computer vision made giant leaps in the past decade. Without doubt, this would not have been possible without the progress in deep learning algorithms. However, these algorithms will soon reach their limits, and thus, researchers are always on the lookout for possible improvements. e unique technology which is ex- tremely promising and which will allow these algorithms to go beyond what we can even imagine is without doubt quan- tum computing. Just to give an example of its immense power, Google managed to use quantum computing on a particu- lar mathematical problem. e system managed to solve it in slightly more than 3 minutes. is result is an immense im- provement considering that the most powerful supercomputer in existence today would take approximately 10,000 years to accomplish the same task! Will artificial intelligence eliminate or increase jobs in the future? It is hard to come up with a ballpark figure of whether AI will eliminate or in- crease jobs. AI experts claim that it will kill more than 40% of all the posts in ex- istence today within the coming decade. If we factor in the effects of the pandemic, this figure is most probably an understate- ment. On the other hand, the World Eco- nomic Forum predicts that AI will create at least 133 million new roles. ese are new jobs which do not even exist today. Furthermore, let us not forget that the IT industry does not operate in isolation and needs many support roles such as visual artists, copywriters and several oth- ers. us, to reach this correct figure, this amount should be at least quadrupled. If we were to aggregate everything togeth- er, the official estimates indicate that AI will most probably create 45% more jobs than it kills. So the future does look bright in general. is prospect does not mean that everything will be plain sailing. Many people will still need to retrain and reskill to thrive in the new job market, but with the help of AI-enabled tools, it will be much simpler than before. Do you believe that artificial intelligence will surpass human intelligence? Intelligence is an attribute which is hard to define. So much so that we do not have a coherent definition for it, yet us humans are extremely good at labelling intelligent beings. Typically, we associate tasks with those that a human can do so if another animal or a machine can replicate those actions, then we label them intelligent. If you take individual tasks such as playing chess, analysing a video feed and many others, AI is not only better than humans but in some instances, it also reached superhuman level. is notion is referred to as narrow AI because it is capable of handling close tasks exceptionally well. However, if you were to ask an AI Chess Grandmaster to relate the current weather in New York, the system fails miserably because the re- quest goes beyond its designed functions. To surpass human intelligence, we need Artificial General Intelligence. is intel- ligence is similar to the human one and can handle multiple different requests simultaneously. Although there has been much research in the past decades, the reality is that we are still very far away. Of course, this might all change with the rise of new algorithms and technologies like quantum computing. AI and the workplace