Get best of the week

Forecasts for 2020 in the field of AI, machine learning and robotic automation


At the end of each year, the eWEEK online publication publishes the opinions of IT idea leaders regarding their expectations for the coming year: new products, innovative services, trends, etc. We bring to your attention a translation of material dedicated to the upcoming 2020. And yes, we remember that it is already March, but these forecasts are still relevant.

Dongyan Wang, vice president of AI transformation at Landing AI


The introduction of AI in the non-consumer sector of the Internet industry is still at an early stage. Many projects are stuck at the pilot level due to difficulties ranging from lack of data to a lack of knowledge on managing complex machine learning processes. In 2020, we will witness the development of end-to-end, vertical AI platforms that will allow customers to withdraw their AI projects from the pilot stage and bring them to the finish line.

Bruce Milne, Pivot 3


The IT market will become aware of the possibilities associated with video processing . In 2020, with the improvement of video analytics, unlimited possibilities will begin to open up for IT. Today, the share of video data reaches 60% of all accumulated information. In previous years, companies regarded the storage of this data only as a duty and additional costs. And this year we will witness a shift: companies will begin to analyze video data in order to optimize their products or reinforce strategic initiatives. For example, with the help of video, cities can not only increase the security of their transport systems, but also implement video analysis techniques to form important conclusions, say, about bandwidth needs.

Matt Kunkel, CEO LogicGate


Robotic Process Automation (RPA) will surpass AI in terms of risk management and regulatory compliance . The reason is that when you need to analyze large amounts of data for Fortune 500 companies, the available information is simply not enough to ensure the relevance of forecasting using AI. The RPA will produce results, since many of the functions related to risk management and regulatory compliance are in line with formal processes. And the more information companies pass through specific processes, the clearer will be how to automate them. The question remains how to optimize and iterate such systems.

Robotic process automation will begin to be applied in areas such as risk management of third parties; IT shaping policies and procedures; internal audit.

David Jones, vice president of marketing, AODocs


AI is not a universal information management solution . We are inclined to consider AI as a universal tool that can solve all the problems of a business by implementing a single algorithm. It's a delusion. It's time to discard the idea that this is the power of one powerful AI algorithm. We need to move on to the concept of numerous AI bots that together optimize previously accumulated data. In 2020, AI will be applied to old databases to find out which data should be left and which can be deleted as unnecessary, and use the enriched metadata to create a better search and simplified storage of records. Not one large algorithm will cope with this, but a set of interconnected algorithms.

Cheryl Wiebe, Practice Lead, Industrial Intelligence Advisor at Teradata


- What the world calls AI in 2020 will be divided into several areas that marketers will probably come up with more meaningful names. This includes robotic process automation; automated selection and development of functionality; Perception AI (perception AI), which automates and improves physical perception; and also AI for resource allocation, combining optimization technologies to perceive and respond to requests in real time.

- AI will begin to improve the data management process itself. For example, from the point of view of the distribution of system resources, the automated design of functionality, the collection of operational metadata and better knowledge management (like tagging).

Jeff Catlin, Lexalytics CEO


Natural language processing and text analytics will play a more important role in RPA solutions . According to Forrester and Gartner, many RPA developers are lagging behind in supporting trends in the use of text analytics. Their solutions lack the ability to “use unstructured documents,” including PDF. And when embedding components of text analytics and natural language processing in various environments, problems arise. As companies automate increasingly large processes, developers of natural language processing tools will offer promising solutions that meet the requirements of RPA: on-premises or hybrid-cloud deployment, easy-to-integrate APIs, customizability, and a quick return on investment.

Chad Meley, Vice President of Marketing, Teradata


- After the appearance of several successful AI pilots in the last couple of years, companies will again focus on corporate data management and integration , which will create the basis for the development of hundreds and thousands of specific ways of using AI. All the varieties of machine intelligence that surround us today are limited artificial intelligence. In 2020, successful corporate AI initiatives will allow the development of hundreds, if not thousands of applications, and a highly specialized algorithm will be created for each of them.

- Great attention will be paid to the creation and implementation of “no code analytics”. We are witnessing a steady process of democratization of advanced analytics through the automation of some time-consuming aspects, such as designing functionality and selecting models. But the real spread of advanced analytics will be facilitated by the development of machine learning and other advanced procedural analytics techniques, when they will require absolutely no programming skills or working with SQL. Analytics without programming will be embedded in workflows or called using simple drop-down menus. This will not lead to obsolescence of programming in the world of analytics, but it will expand hundreds of times the methods of its application in large companies.

Jeff Catlin, Lexalytics CEO


The main achievements in AI research will be theoretical . Over the past five years, the use of AI has far surpassed our understanding of how it works. Given the great practical changes in the second half of 2019, I predict that this year there will be fewer breakthrough developments of algorithms, but we will go further in the theory that explains the operation of machine learning. This area is developing rapidly, and by the end of 2020 the balance will again shift towards a theory that will pave the way for a new generation of algorithms.

Jeff Catlin, Lexalytics CEO


Less magic and more decisions . It will be a good year for AI, it will strengthen its position of defining technology for the next decade. Suppliers are smarter and no longer promote AI as a magical tool. Instead, they correctly say that AI can help people do work faster and better.

Muddu Sudhakar, CEO of Aisera.com


AIOps will destroy traditional IT / clouds / DevOps. At the heart of DevOps is improved responsiveness and flexibility: AIOps can help automate key steps from development to operation, predict operational results and automate responses to changes in the operational environment. Despite the fact that microservices, hybrid clouds, peripheral computing and IoT increase the complexity of applications and increase the volume of logs in which you have to look for the causes of various events, AIOps simplifies the aggregation of data from different systems, while DevOps improves efficiency by integrating previously disparate systems. Like DevOps, AIOps stimulates cultural change, as it requires evaluating the entire system, rather than focusing on specific technologies or infrastructure levels. It also requires a comfortable level with a high degree of automation.

Jeff Catlin, Lexalytics CEO


Self-control ... we have to wait a long time for it . In general, AI will show its best side, but there will be a number of notable failures, for example, in the field of self-driving cars. Smart Summon, Tesla's new model, is very impressive, but it still has a long development to do. The widespread dissemination of this model in the Tesla community will lead to the emergence of numerous videos from accidents at low speed. In these videos, cars will crash into other cars, lampposts and people.

More articles:


All Articles