In recent years, artificial intelligence (AI) has entered the mainstream business domain. According to research conducted by BCG and the MIT Sloan Management Review, AI will have a major impact, in all industries, within the next five years. What companies / industries will AI initially impact?
The issue is what AI really does; it is not a machine or a simple substitution of people for machines. What AI does, which is perceive and see patterns; (this is a glass, you are a lady, etc.) was not possible ten years ago. This kind of technology can be used for a very wide range of applications, most notably voice, face and object recognition, which makes tasks such as automation in call centres, people processes, factories and systematic identification of patterns in data, essentially statistics, possible. So, any company that is in the business of providing products or information services is able to use AI in order to customise them in a way that was not possible before. However, in the case of sales, I do not think I could be persuaded by machines; they are not charming or funny, they do not make eye contact. I am a bit skeptical that a sales function is likely to be taken over by a machine, people are not usually persuded to buy things by machines. With AI you have to set it task by task and therefore you cannot generalise about whether AI will ‘take over’ because there are many different challenges; machines are good at some of these but really bad at others and therefore we will see very different paces of evolution.
Many concerns have been raised around the management and control of AI. Could this be the main threat and is it safe?
No-one knows what will happen in the distant future but the way that a machine learns is from data, it does not learn the way you and I learn. A machine uses data and statistics, it does not understand what it is doing, it does not have common sense – it is purely statistical and therefore the things that machines do is limited by the data the machines has. This means that AI with one set of data is good to solve one set of the problems, it is not about generating general intelligence. So, what we can do is to use machines for those problems where common sense is not needed. We are not yet in a world where machines could replace human beings and they are not a threat to humanity, that is science fiction. It may happen that tomorrow someone will invent a code that may have common sense, but we are not yet even on that trajectory.
In the future AI will take many jobs. Do you think that the use of robots and AI should be taxed?
This argument has been introduced by a number of people, including Bill Gates. In the long run, any kind of automation transforms the work that human beings do and does not create unemployment. The machine cannot replace the entire job, but what it can do is replace specific tasks within jobs and therefore reduce the number of people needed for the job and/or increase the productivity of the people doing the job. There are exceptions – driverless trucks and taxis are coming, but slowly! However when robots replace people for one task, then individuals can focus on another task which creates value because robots are doing all the other tasks and so we all benefit. Therefore, it should not be taxed but rather AI should be encouraged. The key question is how quickly these changes will happen and the speed of adjustment of the economy. This, in my opinion, will happen quite slowly and certainly not overnight. Therefore, my view is that there is no reason to tax it because the economy will absorb the unemployment that it creates.
Digital technologies are reshaping the banking industry. Can you explain how the banking model has been reshaped?
Banking is many different businesses. In many of those businesses technology is working perfectly fine, so there is a lot of exaggeration of how ‘dead’ the banks are. Banks do have some extremely inefficient platforms: international payments, for example, is slow, expensive and it takes time. Similarly, buying a security, it takes a few days for the transaction to be completed because the systems that banks use are extremely old and inefficient. So why don’t the banks have more efficient systems? It is not in their interest! If all the banks are using the same systems, banks have no interest in solving the problem as their only competition is the other banks. But, now Fintechs and movements such as Bitcoin present the possibility of new transaction networks that are outside the banking system and for the first time create a competitive imperative to modernize those antiquated systems.
Cryptocurrencies and the like are not legal tender or regulated…
It depends on the system and the location. With a PayPal account, I can send you the money more quickly and cheaply than via a bank. Alipay has a very high proportion of transactions in China. So, in some context, there already exists a threat to the banking system. It is a really big deal in poorer countries – Africa, South America etc., when the banking system only serves a fraction of the population and the alternative payemnt systems can go directly to the masses, making it possible to make payments even with a digital currency. What we see now is that the banks are waking up and realising that their competitor is not another bank, but PayPal, Bitcoin, Alipay, etc.
But these threats are from the gig economy and are not fully legalised, so what about safety?
The answer is different from place to place; maybe they are legal in one country and not in another. You have to ask about the risks of the new systems and the problems they solve. There is no universal answer. Paying with Bitcoin is not illegal but the question is how it should be or whether it should be regulated? Bitcoin serves two purposes: it is a medium of exchange and a store of value – I can buy a Bitcoin in the hope that its value will go up. In recent years, Bitcoin has been more important as a store of value, an asset, transactions with Bitcoin are growing slowly – 10% – 12% a year.
Do you think central banks will invent their own cryptocurrency?
Central banks have good reasons to introduce their own digital currency; a ‘Bitdollar’ or a ‘Biteuro.’ There are also a lot of questions around its design, the problems it will solve and the platform that will make it possible. A central bank, for example the European Central Bank (ECB), could create its own digital currency and everyone could have an App that enables payment with euros settled through a blockchain –thousands of computers that all record a transaction. The problem is that blockchain technology is currently very inefficient. The Visa system processes around 2,000 transactions a second, Bitcoin processes about five transactions a second and it cannot go much faster. But, despite the inherent inefficiency of blockchain technology, it will not go away. You could set up a digital technology but in a totally different way, for example PayPal works with a single server and when I pay you, a single message goes through and the transaction is completed in our respective accounts. However, we have to trust PayPal. With Bitcoin you do not need to trust anyone. So, the central banks could set up “Eurocoin”, just like a Bitcoin, as a decentralised system that is digitalised but inefficient. Or they could simply allow everybody to have an account