LATEST TRENDS OF ARTIFICIAL INTELLIGENCE

Artificial intelligence will revolutionize industries, whether to replace man in his grueling tasks with automation or to help him in mechanical or medical diagnosis.

Over the years, companies have been training their algorithms to make them more efficient. Their progress is due in particular to the evolution of the electronics offering them ever larger computational powers, or optimized systems to be able to make the algorithms run closer to the object. Among the many developments, here are the latest trends that should jostle 2019 and 2020 beyond.

Artificial intelligence chips

Artificial intelligence requires a large amount of data and, therefore, a great deal of computing power to achieve learning by reinforcement (deep learning). Electronics manufacturers have developed chips dedicated to supercomputers, as is the case of Nvidia, Intel or AMD, but also inference chips for data centers to quickly run deep learning algorithms préentraînés. The goal is to have an artificial intelligence deported (on the cloud) more efficient. Qualcomm has just launched its Qualcomm Cloud IA 100 chip this year. Web giants such as Facebook (partnership with Intel), Amazon Web Services or Google Cloud are also on the spot.

Embedded intelligence

In cyberphysics systems based on artificial intelligence, the issue of embedding computing power or deporting it to the cloud comes back regularly. For the sake of connectivity or data security, companies have developed embedded artificial intelligence. This AI "on the edge" performs the calculations and its learning on the object thanks to the optimization of the algorithms so as not to require high computing power. This is particularly the case in industry for artificial intelligence startup Halfcode or Amiral Technologies, proposing to do predictive maintenance.

medical diagnostic aid

One of the areas in which artificial intelligence could have a major impact is medicine, especially in helping cancer diagnosis. Google recently collaborated with Northwestern University Hospital to develop an AI that can detect lung cancer. The web giant had already demonstrated its expertise in this area with the detection of breast cancer. Other research focuses on different types of cancer, such as the cervix.

Facial recognition

Facial recognition could take a new turn in 2020. Installed in recent years in China, it was the subject of an experiment in February in Nice during the Carnival with 1000 volunteers who gave their consent to be targeted by the cameras. The goal was to test the access control, the detection of an individual in a crowd or a person of interest on the public road. Amazon has also been the subject of claims for the sale of its Recognition technology to government agencies.

Smart assistants

We remember last year, the demonstration of Google and his voice assistant "Duplex" who makes a restaurant reservation punctuated with "uh" and "hum". This month, the New York Times revealed that the entire service was not performed by this assistant and that humans were behind the calls received in 25% of cases. New projects, such as Mozilla's Common Voice, are launching to compete with the industry's best-known names such as Alexa or Siri. Progress is expected in the field.

MACHINES THAT UNDERSTAND THE CAUSAL LINKS

Today, when algorithms, such as chatbots or virtual assistants, respond to our requests, this is because they are programmed by scenarios defined in advance. The machine only executes what is asked of it without understanding the causal link. Unlike humans who, through common sense, learn the causes and effects over time. In 2020, we will see the algorithms understand that. Machines will now be able to link demand and the resulting action with increasingly diversified data sources from the growing number of data collected and better exploitation of data.

4TH INDUSTRIAL REVOLUTION - DEMYSTIFICATION OF AI

Although AI has gradually become a part of everyday life, it continues to worry. It raises particular fears because of its impact on society and employment. This is more of a belief than observation and rational analysis. In fact, it is rather the companies that will not adopt this technology that will suffer significant economic consequences. Competing with organizations using AI, they will no longer be competitive and will ultimately be in danger. For the latter, 2020 will therefore be a pivotal year: hence the importance for them of informing and educating their employees on the benefits of this innovation. We are at the dawn of a new technological revolution, remapping the cards between all the economic actors.

AI TO IMPROVE ITS DIGITAL TRANSFORMATION

Although digitization is only the first step in a digital transformation process, it is nevertheless crucial. It allows companies to collect data that will feed into other technologies and, ultimately, improve their performance. Thus, to be profitable, digitization only makes sense if it is coupled with other tools. By 2020, it is likely that many digitization projects will be associated with initiatives to improve the customer experience or to increase operational efficiency in order to increase revenues. Artificial intelligence can help businesses achieve these goals.