When machines imitate humans
Artificial Intelligence refers to a set of technologies – machine learning, deep learning, language processing, etc. – that share one common feature in that they rely on a computer system capable of analyzing, understanding, learning and discovering connections between things, facts and events as well as manipulating concepts. It should come as no surprise that machines have acquired these extraordinary abilities. Just like flying cars, autonomous and hyper-intelligent humanoid robots have been a major part of science fiction for decades.
“Artificial Intelligence is a word that has been around for 60 years, but which ultimately refers to nothing more than software. Machines are very good at performing repetitive tasks and can help humans work more efficiently. But they cannot take their own initiatives and can only make progress by interacting with people”, explains Edouard d’Archimbaud, manager of the Data Science & Artificial Intelligence Lab at BNP Paribas CIB.
And yet, because of the fear that this highly efficient software will eliminate jobs and concerns about the power Artificial Intelligence may wield over people, we still have a distorted, often catastrophic image of AI.
“ Artificial Intelligence is a word that has been around for 60 years, but which ultimately refers to nothing more than software. Machines are very good at performing repetitive tasks and can help humans work more efficiently. But they cannot take their own initiatives and can only make progress by interacting with people ”
Manager of the Data Science & Artificial Intelligence Lab at BNP Paribas CIB.
Ingenious computer programs
One of the creators of Artificial Intelligence, Marvin Lee Minsky, notably defines it as “the construction of computer programs that engage in tasks that are, for now, more satisfactorily accomplished by humans because they require high-level mental processes”.
Today, Artificial Intelligence is in fact likely to surpass humans in performing tasks that require reasoning and learning. Machines now have the ability to understand human language, grasp the meaning of a sentence and deliver an adequate response.
Already in 2011, IBM’s computer Watson won the trivia game Jeopardy; the machine had to understand the question before finding the answer!
In 2016, Google’s program AlphaGo beat the South Korean champion of Go, a game widely seen as the last arena in which humans could still outdo machines… All these feats of artificial prowess fascinate and sometimes worry us. In this way, we are no different than the people of the 19th century watching the first steam engines pull into a station!
A massive impact on daily life… and the economy
Artificial Intelligence represents genuine disruptive technology, which will have an even greater impact on our daily life than computers, the Internet or even the smartphone. The effect it will have is still often underestimated today. And yet, as Accenture revealed in “Why Artificial Intelligence is the Future of Growth”, Artificial Intelligence could boost productivity in France by 20% by 2035, driving annual growth up from 1.7% of GDP to 2.9%.
AI has the potential to weave its way into every field of business – producing smarter machines for industry, revolutionizing education and research, reinventing points of sale and medicine as well as entering our day-to-day lives in the form of smart, connected objects. By optimizing the supply chain, helping to streamline trade and reducing demand for energy, AI could even help combat global warming.
However, the tremendous potential of Artificial Intelligence requires extreme caution, we must implement ethical rules and strong regulation (just like any technology revolution). Amazon, Google, Facebook, IBM and Microsoft have already joined forces to form an alliance to share progress and reflect on the potential dangers of their discoveries: Partnership on AI.* * The full name of this alliance is “Partnership on AI to Benefit People and Society”.
Toward reinvented, personalized banking services
Banking and financial sectors will also be able to take advantage of Artificial Intelligence to offer their customers innovative and personalized services as part of a more streamlined user experience. BNP Paribas is preparing for these challenges by forming research teams and fostering close relationships with a wide range of players in this booming industry. For example, BNP Paribas hosts startups focusing on Artificial Intelligence within its Fintech accelerator, and also takes part in fundraising initiatives launched by innovative companies in the field.
As head of the AI Lab at BNP Paribas, Edouard d’Archimbaud is striving to achieve two main ambitions, “The first is to automate every task that will help make the bank ‘scalable’. The second is to use data to offer innovative services to our customers, services that go beyond the current banking model and contribute even more value-added to the services provided by our employees.”
Within the Lab, data scientists, developers and web designers do not stop at exploring the potential of Artificial Intelligence. Their developments are put into production in the form of APIs. Artificial Intelligence has already led to practical applications.
Among them, an automated system for analyzing contracts that makes it possible to “read” and understand contracts up to 150 pages long in just seconds. For employees tasked with ensuring the compliance of these documents, that saves an enormous amount of time! Building on deep learning, Edouard d’Archimbaud’s teams have also developed an automated translator with unparalleled performance. Extremely promising work on voice recognition is also underway to create an API that can transform voice into text in a fluent and faultless way. This is an added challenge that only human intelligence can achieve – with the help of machines, of course.
The lingo of AI
Machine learning refers to an AI system’s ability to learn “by example” or “by experience”.
Deep learning is a learning technology that uses artificial neural networks, which approximate human learning to process “raw data”. For example, deep learning allows a machine to identify what an image represents, or recognize a single word in any language, etc.
Language processing refers to a machine’s ability to understand and process natural language. This is an important concept in AI because it relates to the “Turing test”, developed in 1950 by Alan Turing as a criterion for intelligence (the test evaluates a machine’s capacity to imitate human responses within a real-time, text-based conversation in which the human participant is unable to distinguish the machine from another human being).