What to expect from AI in 2019: generative networks and “vague calculations”
Every year, AI -solutions appear in more and more industries – to ensure the safety and diagnosis of diseases, translation of texts and hiring of employees, in agriculture and industry. The possibilities of technology are not limited.
AI expert, venture capital investor Kai-fu Lee, in an interview with CBS, prediced that in the next 15–20 years, artificial intelligence would leave about 40% of people out of work, including taxi drivers, waiters and even chefs.
According to the forecast by Artificial Intelligence Market by Technology and industry vertical , the average annual market growth rate in the period from 2018 to 2025 will exceed 55%. Consider what will bring us the AI-solutions in the coming year.
Computer vision will be used in production
This technology is being developed thanks to commercial interest in unmanned vehicles, security systems and monitoring. The main directions are object detection, object segmentation, object tracking and object tracking. An example of computer vision projects can be the German company Nyris , whose algorithms are improved with each new image. Nyris uses several technologies to analyze objects in photos and videos, providing a result within a second. In 2016, the project received support from the Google Startup Program.
In 2019, in the field of computer vision, the developments that have already appeared will be improved.
Now the technology is developing at the hardware level. New sensors better recognize surfaces of different types and build three-dimensional models of objects. This optimizes manufacturing processes.
Development of machine translation and text tonality analysis
Natural language processing solutions are rapidly progressing. In some areas of the translation machine is already comparable with people. In particular, it concerns medical texts.
It is important for representatives of business or politics to know what their image is in the information field – what is written about them and how. In addition, companies can use text analysis to communicate with customers. Such analysis can also be performed by neural networks. The cogito project , which in the summer of 2018 received investment in the C series, helps support staff understand the mood of the client who approached them.
In the analytics of English texts, leadership is still maintained by Facebook and Google. But small and medium businesses have the opportunity to occupy a niche of working with more rare languages that are less interesting for large companies.
Peripheral computing will begin to crowd out cloud
Peripheral (or “foggy”) computing is an analogue of cloud computing, in which data remains within the local network. Perform foggy computing will be computers inside smart home devices or local cloud servers. The advantages of peripheral computing are that they are cheaper, less dependent on the stable operation of the Internet, reduce the load on Internet channels and reduce the response time. In addition, the placement of data on the local network will increase their security.
AI will change healthcare
According to a study by CB Insights the state of Artificial Intelligence 2018 , in 2017, it was AI-startups in the health sector who received the largest amounts of funding in the market. The trend persists. Companies are developing wearable gadgets for monitoring, as well as solutions in the field of disease diagnosis, work with patients, clinical research and other areas.
In 2018, the cloud-based platform for the diagnosis of cardiac and oncological diseases Arterys has achieved remarkable success . The company has received official approval from the Food and Drug Administration (FDA) to apply technology that helps doctors diagnose lung and liver cancer. In 2017, the previous development of Arterys for cardiac diagnostics was also officially approved by the FDA.
In January 2019, researchers reported significant breakthroughs in the use of neural networks in medicine – an algorithm learned to detect cervical cancer from a photograph with an accuracy of 91% (20% more accurate than doctors). Three groups of other scientists, according to Science, spoke about the successful decoding of brain activity to speech — previously, the signals that the brain of a paralyzed person could transmit could not be transformed into text.
Models based on deep learning have not yet occupied a leading place in this area only because of its conservatism. In most countries, it is forbidden to use for the treatment of patients technology, the algorithm of actions of which is opaque (as is the case with neural networks). It is not clear how to interpret the choice of network in case of an error.
Computer games will be more realistic
This will happen thanks to the generative networks (neural networks whose task is not to recognize the image or sound, but to create it). They will also be used for 3D modeling, voice assistants enhancements, photo enhancements, or any images. Generative neural networks underlie the Reflect application developed by the Ukrainian startup NeoCortext. Now it allows you to replace the faces in the photos. In the spring of 2019, users will be able to work with GIF and video. Later, the body replacement option will appear.
People will strive for full transparency of AI
The public cannot trust technology until it understands how it works. But the principles of the neural networks – what is happening in the “black box” is still a mystery. In 2019, AI will affect all areas of life. Therefore, the need to find out how the neural network works will become a matter of principle.
In 2019, Artificial Intelligence will produce real results in many areas, from image generation to healthcare. People will use AI-based gadgets more than ever. For example, more than 66 million Americans will resort to voice assistants, according to an eMarketer study. The penetration of Artificial Intelligence in different areas of human life will lead to the fact that the AI component in the product will no longer be a competitive advantage. Therefore, we at GR Capital will more thoroughly analyze the new proposals with AI at the base. We believe this trend will be relevant for the entire investment market.
This site uses Akismet to reduce spam. Learn how your comment data is processed.
Recent Posts: TechnoBlogy