The basics of AI appear to be well established and unlikely to make major changes in the short to medium term.. Focus on the improvement and variations of existing technologies. .
According to a top 15 made by Linkedin, “artificial intelligence engineer” will be the second most sought-after profession in France in 2020, especially in info-com, computing and the l’info Web. The sector is booming. Finland,which was at the head of the European Union until 31 December, has announced that it wants to introduce 1% of European citizens to AI free of charge, so that the Old Continent will be better equipped in the future. But for what functions?
In his blog post, 20 Predictions and 0 Bullshit, AI consultant and teacher Eytan Messika argues that two major areas will grow in 2020: those related to basic research – the basics in other words – and those related to functions peripheral to new applications. The development of autonomous cars, he writes as an illustration, leads to “the creation of microservices such as vision software (Mobileye), sensors (Velodyne), simulation platforms (Cognata) to support this new market”.” In the same spirit, data protection and ethics services would be expected to expand in 2020.
In general, the air of the time is to en implement. “The fundamentals of AI should not change or little,” explains AI scientist Jean-Claude Heudin. Much of the innovation will come from the development or declination of existing technologies.” According to Renaud Champion, director of the New Intelligence division at EM Lyon Business School, everything related to human-machine interactions is already accelerating. “Chatbots -t-il are adopted everywhere,” noteshe, on the websites of banks, insurers, mutuals, in reception halls, train stations,etc.”
These conversational agents allow you to quickly give some simple information to potential customers. But don’t expect them to really have discussion! “It still seems a long way off, the current issue is more on the side of the customer relationship”, argues the specialist. Software capable of quickly giving an appropriate answer, to a person’s questions or behavior, can trigger a purchase, a subscription or even a subscription.
“Cobotic”and diagnostic aid
“Relatedly, the collaborative robot – also referred to as “cobotics”- is exploding in the industry,” continues Renaud Champion. Before, the tendency was to use on one side large machines, as on assembly lines, and on the other supervisory personnel, for everything related to logistics,for example.. Today,things seem less cleavage.” Machines of all sizes, packed with sensors and more or less autonomous systems, assist staff at all levels: technicians, operators, production managers…
Anything that optimizes the flow of goods is popular. “The main limitation to these technologies,” says Renaud Champion, “is gripping” is to grasp something. Indeed, machines do not have the flexibility and adaptability of human hands, which limits their deployment in certain sectors of agri-food, for example. “However,”he continues, “all industrial applications to reduce work hardship are expected to develop in the short to medium term.”
Declinations are also multiplying in the health sector. Everything related to image analysis – radios, scanners, etc. – and therefore diagnostic aid, benefits from the progress of deep learning in recent years. “Ai-assisted medicine is becoming democratized,” adds Renaud Champion, pointing to the rise of surgical robotics. Surgeons use small AI systems to prepare an operation, simulate a gesture or refine their own practice.”
Manufacturers should also develop tools to shorten the duration of examinations, to support patients more effectively and to better manage patient flows. Some hospitals are considering using AI to diagnose “bobology”at the entrance to emergency departments, or to monitor patients remotely through automated alert systems – and thus relieve congestion in services… For Jean-Claude Heudin, the human being is still indispensable. “The clinical validation of these kinds of tools is not for tomorrow, it will take a lot of time for this to be fully integrated into a care protocol, if at all possible one day.”
More generally,”many issues will revolve around the integration of these technologies into our work or our daily life”, notes this specialist. The autonomous car, for example, is in itself almost ready.. But the urban network is not configured to accommodate their traffic: a technological infrastructure, but also cultural, will have to be designed and democratized in the years to come.. This is the whole problem of the smart city: systems for capturing, analyzing and transmitting urban data should be developed to allow AI to better interact with their environment..
Similarly,”in companies, I am regularly asked to facilitate the acculturation of executives,” says Jean-Claude Heudin.. According to him, the AI today is a bit like the Internet or digital yesterday: all the boxes will use it tomorrow to varying degrees, sometimes in a very basic way and without necessarily making it a business in its own right. “In personnel management, for en example, it is clear that we are going to use more and more partially automated tools of matching, training or mobility.” Humans will retain a leading role, but will be able to offload the most repetitive tasks and focus on the higher-level functions. “It will take time, because it is also a cultural upheaval. But everyone will come,” concludes Jean-Claude Heudin..