Artificial Intelligence Technologies: Four Waves of Development

The first artificial intelligence technologies appeared more than 50 years ago, but only recently this area began to actively develop. Although modern algorithms are still very far from the classical in our understanding of AI, they are already capable of solving many problems in medicine, business, science and other areas. Ultimately, autonomous systems will surpass humans. In the article, we examined the basic principles of artificial intelligence, the stages and directions of development, as well as the further impact on the economy, industry and humanity..

The essence and principles of AI

Term «Artificial Intelligence» (AI) first sounded in 1956 at the Dartmouth conference, but only in the last decade has it begun to develop actively. We are now seeing only the first significant glimpses of its potential and applications. The main task of AI is to perform tasks originally intended for human cognition: pattern recognition, forecasting and making complex decisions..

Algorithms can perceive and interpret the world around us, and some believe that they will soon be able to feel emotions, compassion and create. Although the dream of an AI completely identical to the human mind is still very distant.

The turning point in the development of computer technology was «deep learning» (deep learning) – architecture based on the principles of the functioning of the human brain, neurons and connections between them. Such systems can consist of thousands of layers, billions of parameters, and analyze a huge amount of variable data. However, unlike the human brain, their development is based on mathematical choice and pattern recognition from data arrays..

Any type of information system can be used to enter the initial parameters: images, sound segments or credit card payments. After processing them, artificial intelligence provides a solution or forecast. For example, what words are spoken on the recording, whether the transaction is fraudulent or further market behavior.

The breakthrough was facilitated by «data explosion», coming from the Internet and related to activities, intentions, preferences of people. While the human mind focuses on the most obvious connections between inputs and outputs, machine learning algorithms, by analyzing the sea of ​​data (big data), can reveal such subtle correlations that we cannot even describe them logically..

Combining hundreds or thousands of layers provides an edge over the most experienced experts. AI algorithms are now superior to humans in speech and face recognition, various games, and the diagnosis of certain types of cancer based on MRI results..

However, for analysis, they need a ton of data to train and huge computational power to process it all. Modern AI also functions only in specific areas and is not able to generalize or use common sense..  For instance:

  • AlphaGo, who defeats world champions in the ancient game of Go, doesn’t play chess;
  • algorithms trained to define loan underwriting fail to allocate assets.

Despite the existing limitations, under the influence of computer technology, artificial intelligence has moved from the era of discovery to the era of implementation. At the moment, the center of gravity has already moved from research laboratories to real applications. Companies and governments are actively exploring this area, looking for ways to use modern AI capabilities in their operations in order to squeeze the maximum performance out of this innovative technology. The potential of artificial intelligence increases and the fact that it can be applied in almost any field.

Four waves of AI development

Artificial intelligence technologies can be divided into four waves, which able to exist in parallel, but with different starting points and rates of development.

The first stage can be called «Internet AI». His work is based on a huge amount of information located on the web. The algorithm uses the fact that users automatically flag the data in the process of viewing it: make purchases or not, click or ignore. These labeled data arrays form a detailed profile of our personalities, habits, requests and desires: the perfect way to compose more personalized content, keep consumers on a specific platform, maximize profits..

Second wave – «business AI». Algorithms can be trained on private datasets, from customer purchases to machine service records and complex business processes. This ultimately allows managers to improve their decision-making process. For example, the system can study thousands of bank loans and repayment rates, revealing which type of borrower has a hidden risk of default or, conversely, lost clients with excellent prospects..

Medicine can also use deep learning. Processing a huge amount of data on diagnoses, genomic profiles, therapy results and subsequent health conditions will help develop a treatment algorithm for each individual patient.

Artificial Intelligence Technologies: Four Waves of Development

The third wave of artificial intelligence can be called «AI perception». He will be able to see, hear, and have other senses on his own. This will allow him to collect data that has not been previously recorded and use it to create new applications. Already, many smart devices are starting to connect our homes and cities to the Internet of Things. We are on the verge of entering an economy with trillions of sensors that include voice interfaces, computer vision. Digital scents are also being developed and ways of transmitting other sensations over the Internet.

The fourth wave is monumental and the most difficult: «autonomous AI». By integrating all the previous directions, artificial intelligence will give mechanisms the ability to feel and react to the world around them, intuitively move and manipulate objects as easily as a person does.. 

This wave includes completely autonomous devices that will be able to see the world around them: recognize patterns in the pixels of the camera, figure out what they relate to and then use that information to make decisions. In the field of robotics, such advanced AI algorithms will be applied to:

  • industrial applications (automated assembly lines and warehouses);
  • commercial tasks (robots for washing dishes and harvesting fruits);
  • and, ultimately, consumers.

Further development prospects

Algorithm programmability allows you to maximize profits or replace human labor. This increases the economic value of artificial intelligence technology. AI is fast, accurate, works around the clock, does not complain and can be used to solve many problems with high efficiency.

PwC, one of the largest consulting firms in the world, estimates that the technology will create a $ 16 trillion market by 2030. AI will be able to replace not only administrative, but also technical personnel, workers.

While there is a threat that algorithms will reach superhuman levels and potentially threaten our existence, this is still a long way off. To form such «absolute AI» he will have to have many possibilities:

  • reasoning;
  • conceptual learning;
  • common sense;
  • planning;
  • abstract thinking;
  • creation;
  • self-awareness and emotions that remain beyond our understanding.

However, up to this level, dozens of major scientific breakthroughs will be required. Over the past 60 years, only one has been completed, so you don’t have to worry about this problem..

The AI Technology Revolution

However, even autonomous AI will drive a new industrial revolution and other global changes. Therefore, we need to understand its tremendous impact, long-term benefits and existing challenges..

Virtual reality technologies, which will significantly change many professions and jobs, will also play a significant role in the world of the future..

text: Ilya Bauer, photo: Unsplash, cdna2.artstation

Similar articles