Recently, a report about companies misusing artificial intelligence for propaganda came out. According to the report, in Europe, 40% of startups claiming to use AI do not use the technology.
Last year, some research institutions accidentally discovered that some companies violated the rights of users. They claim to use machine learning and advanced artificial intelligence to collect and check the personal data of thousands of users in order to enhance the user experience of their products and services.
Unfortunately, the public and the media are still confused about the real “artificial intelligence” and the exact definition of “machine learning”. Usually, these terms are used as synonyms; in other cases, the two technologies have different meanings. People with ulterior motives use the blind spots of the public to hype up gimmicks for profit.
Machine learning is not the same as AI
What is machine learning? To quote Tom Michel, the interim dean of the School of Computer Science at Carnegie Mellon University and a professor in the Department of Machine Learning: A field of science is best defined by the central problem of its research. The field of machine learning tries to answer the following questions: “How do we build a computer system that automatically improves with experience, and what are the basic laws of all learning processes?”
/ Artificial intelligence includes a large number of technological advances that we are already familiar with, machine learning is just one of them. /
Machine learning is a branch of artificial intelligence. As Professor Michelle defined: “Machine learning is the study of computer algorithms. These algorithms allow computer programs to automatically improve through operating experience.” Machine learning is one of the methods we hope to achieve artificial intelligence. Machine learning relies on examining and comparing data to find common patterns and exploring nuances to deal with data sets of different sizes.
A common type of machine learning is supervised learning. The algorithm will try to predict the relationship between the target output and input features, and build a model based on the relationship, so as to predict the output value of the new data based on the relationship between these data. The formulas used for prediction are obtained from the original data set.
Unsupervised learning is another type of machine learning. In the family of machine learning algorithms, it is mainly used for pattern detection and descriptive modeling. Reinforcement learning is the third mainstream type of machine learning, which aims to use the observations collected from the interaction with its environment to take actions to maximize benefits or minimize risks. A typical example of reinforcement learning is to make a computer’s IQ surpass humans and defeat real humans in computer games.
The many types of machine learning can be dizzying. Especially its advanced sub-branch, such as deep learning and various types of neural networks. Although some people tend to compare deep learning, neural networks, and the way the human brain works, there are essential differences between the two.
And artificial intelligence covers a wide range. According to Andrew Moore, the former dean of the School of Computer Science at Carnegie Mellon University, “Artificial intelligence is the science and engineering that enables computers to operate with human intelligence.”
It is very appropriate to define artificial intelligence in such a sentence. However, it shows how broad and ambiguous the field is. Fifty years ago, the program of playing chess was considered a form of artificial intelligence. Because game theory and game strategy are functions that only the human brain can perform. Nowadays, chess has become a small game that comes with almost every computer operating system.
Artificial intelligence includes a large number of technological advances that we are already familiar with, and machine learning is just one of them. Past artificial intelligence work results have used different technologies. For example, the artificial intelligence “Deep Blue”, which defeated the human chess champion in 1997, used a method called “tree search algorithm” to evaluate the millions of moves that may occur in each round.
In March 2019, three scientists engaged in deep learning research won the Turing Award. From left: Yann LeCun, Geoffrey Hinton, Yoshua Bengio
As we all know, with the emergence of Google Home, Apple Siri, and Amazon Alexa, artificial intelligence has become a symbol of human-computer interaction tools, and the video prediction recommendation system supported by machine learning provides power for Netflix, Amazon and YouTube. These technological advancements are gradually becoming an indispensable part of our daily lives.
Contrary to machine learning, the goal of artificial intelligence is constantly improving, and its definition continues to evolve with the progress of related technologies. Maybe in a few decades, today’s innovative artificial intelligence advancements will be considered as boring as flip phones.
Keep your distance from “artificial intelligence”
The artificial intelligence industry has experienced many twists and turns. In the first few decades, the entire industry was hyped. Many scientists believe that the emergence of human-level artificial intelligence is just around the corner. However, the unfulfilled assertions have disappointed the entire industry and the public in artificial intelligence, and led to a cold winter in the industry. The external financial support and interest in this field has been greatly reduced.
At the time, artificial intelligence had become synonymous with hype. Some companies want to distance themselves from the term “artificial intelligence” and use different names to refer to their work. IBM described Deep Blue as a supercomputer and denied it even if it did use artificial intelligence.
During this period, various other terms, such as big data, predictive analytics, and machine learning, began to gain attention and popularity. In 2012, machine learning, deep learning, and neural networks made great progress and were applied in more and more fields. Many technology companies have also started to use the terms “machine learning” and “deep learning” to promote their products.
Deep learning is mainly used in fields such as speech and facial recognition, image classification, and natural language processing. Originally, these fields were in the early stages, but they suddenly developed by leaps and bounds after the advent of deep learning. In March 2019, three scientists engaged in deep learning research won the Turing Award. These achievements make deep neural networks a key component of today’s computing functions.
Therefore, with the rapid development of other technologies, artificial intelligence is also ushering in a new spring. Today, even in non-profit organizations, machine learning and deep learning engineers can earn high salaries. This fully shows how popular the field is.
Sadly, some external propaganda has overhyped and misreported this technology boom, and even often included pictures with crystal balls or pictures of other supernatural phenomena in articles about artificial intelligence. This kind of ignorance will only help companies that abuse artificial intelligence to promote their products.
However, on the way of technological development, because they cannot meet public expectations, these companies have to hire manpower to make up for the so-called artificial intelligence. Finally, they may once again lead to public distrust in the field, and because of short-term benefits, another artificial intelligence winter will come.