Prospects for Artificial Intelligence in 2020

Less marked data, more talking devices

It is known that in order to obtain a useful result from ML, a larger volume of initial data is required. Faced with the problem of the lack of suitable data for processing, specialists use various methods of artificially generating new data based on the source. However, there is a converse problem, which will become more relevant every year: the problem of excess data.

The fact is that in connection with the development of the Internet of Things, the exponential growth of connected devices is forecasted for the coming years. The speed of data generation far exceeds the growth rate of computing power already. The "narrow neck", however, is in another place: human resources are insufficient to mark out all the available data. Each month, the index of marked data falls among all the digital tracks left by people and cars. This means that the options of “learning with a teacher” are becoming irrelevant.

In this regard, the anticipated development of the technology of "learning without a teacher" is forecasted — GAN-technologies, ML on streaming data, correlation analysis, reinforced learning, etc. There is another forecast related to the rapid increase in the number of connected devices. Experts expect that in the coming years, many such devices, especially from households, will begin to install autonomous speech recognition and synthesis systems for voice control.

Computer vision enters everyday life

Computer vision technologies are no longer laboratory exotics and are increasingly becoming familiar. In retail, the automatic formation of analytics of visitors moving around the hall is introduced — this allows more efficient management of the placement of goods on the shelves.

Systems of automatic analysis of the input flow of people also appear, which makes it possible to flexibly control the number of working cash desks at the exit. In large organizations, automatic face identification access for employees e is introduced — both in the premises and in the computer hardware. The latter eliminates the necessity for regularly inventing and changing access passwords. During the year, analysts also expect the introduction of a face pass at the airports, which will exclude delays in registration and passport control.

In the next year or two, a second wave of QR codes is foreseen: they will be primarily used as more convenient means of transferring money compared to phone numbers. This practice has already become dominant in China, for example.

Speech and text analytics for business and home

In large companies, where a lot of historical data has been accumulated for training, some types of documents are already prepared and analyzed by robots better than by a lawyer with average qualifications. In the corporate sector, voice-based text generation services are increasingly being used — letters to technical support, equipment orders, travel arrangements, etc. Over the next 2 years, penetration of virtual voice assistants into the management of home appliances is also projected.

It is also clear that the coronavirus pandemic will significantly affect the development of the industry this year. In particular, the world will think about how to effectively distance themselves in such difficult times and how AI could help with this issue.

Image courtesy of Forbes