Technology optimists see Artificial intelligence (AI) as the solution to most economic and social problems of the 21st century and beyond. While doomsayers caution about AI takeover and propose early preventive measures to put thinking machines under strict control. Let’s dig into the AI backgrounds and answer the most frequently asked questions about it.
How AI industry has been so far?
AI growth has been exponential and has followed Moore’s law of computation power, doubling every two years. There has been an explosion of training data available, and after certain hardware capacity thresholds were crossed, previously theoretical algorithms became practically feasible. For example, the IBM Deep Blue project started in the 1980s, and after many unsuccessful attempts it finally beat a human chess grandmaster in 1997.
Predictions about long-term AI growth and its economic impact may resemble daydreams. However, Moore’s law has held true for 120 years. The smartphones in our pockets today are more powerful than the supercomputers of the mid-20th century, although arguably few people could have imagined such devices in the 1980s. Ray Kurzweil, a futurist and AI scientist at Google, proposed the Law of Accelerating Returns, which states that technological growth accelerates over time and that the growth rate increases due to accumulated knowledge. If this law holds as well, AI will not only continue expanding exponentially but could also significantly accelerate the growth rate.
While machine wins over humans in games such as chess and Go or in tasks such as image classification are spectacular, these are examples of Artificial Narrow Intelligence (ANI) for well-defined tasks. The dream of human-level Artificial General Intelligence (AGI) remains in the distant future.
The fearmongers of AI usually refer to Artificial Super Intelligence (ASI), the next evolution step after AGI. They are afraid AGI will teach itself so fast that this will lead to intelligence explosion far above the levels humans could even imagine now and control in the future.
At the same time Andrew Ng, a well-known AI researcher and practitioner, “doesn’t see a clear path for AI to surpass human-level intelligence” without some new technological breakthrough.
Andrew Ng talk at Stanford GSB
Facebook’s CEO Mark Zuckerberg supports implementation and dissemination of AI. He is AI optimistic, Zuckerberg believes that “In the next five to ten years AI is going to deliver so many improvements in the quality of our lives.”
Mark Zuckerberg online Q&A session
Elon Musk has an opposite position. He has said AI is “the greatest risk we face as a civilization”. Elon Mask founded a non-profit company OpenAI, whose goal is “discovering and enacting the path to safe artificial general intelligence.”
Elon Musk — With artificial Intelligence we’re summoning the Demon
The discussion of AI’s distant future and its impact on humanity is wider and deeper than tech billionaires’ tweets or smart people quotes. For a better understanding principles of AI development, we highly recommend the deep and entertaining two-part series “The AI Revolution: Our Immortality or Extinction” by Tim Urban in his “Wait But Why” blog whose big fan is Elon Musk as a starting point that offers many links to other sources and research.
We are AI optimists and agree with the people who say that AI will bring benefits and will be the linchpin of the fourth industrial revolution. AI will not only boost productivity and create new large industries, it also will lead to breakthroughs in healthcare, biology, climate research and many other fields. AI will displace many jobs, but it will also create many new opportunities and employment models.
However, potential job displacement by AI is a serious problem for many occupations. A recent study of Carl Benedikt Frey, Michael Osborne “The future of employment: How Susceptible are Jobs to Computerisation?” found that 47 percent of total US employment is in the high-risk category for potential automation over the next two decades.
You could check if AI will automate your job on the site willrobotstakemyjob.com.
AI is actively used by everybody. Many people speak with Siri, which uses AI technologies. The largest search systems, like Google, Baidu and Yandex, use AI in search algorithms. AI helps to rank pages by semantic matching rather than ranking pages by word search. AI is also used in Google Translator; as a result, the AI text translator becomes more similar to human speech over time. AI algorithms have been used by image search for many years. Google DeepMind, specializing in AI research, has trained AI not just to find images but also to create absolutely new images based on a text.
Nowadays, the vast majority of tech giants actively implement AI solutions. The smartest AI assistant was developed by Google and has an IQ score of 47.28. The second smartest is Baidu (32.92), and the third is Bing from Microsoft (31.98), while Siri’s score is 23.94. In 2014, Google’s IQ score was 26.5, while Microsoft’s was 13.5. These improvements show the significant progress made by AI researchers at tech giants. For comparison, the average IQ score of a 6-year-old child is 55.5, currently unachievable for tech-giant AIs. However, using an IQ-based approach to assessing AI is limited: An AI can be the best in one field — for example, chess or AlphaGO — but can be ineffective in others.
AI is created not only by tech giants. For instance, an android developed by the United States Army has been trained to cook by watching videos. It has learned to cook from visual recognition and trial and error. Cooking uses a wide range of skills and demonstrates the robot’s impressive abilities.
You may ask a question: Why and how does AI develop? Large datasets, not algorithms, are the basis for AI development. For instance, Google uses reCAPTCHA, which protects websites from bots, to assist in the digitization of books and labeling of images. As a result, reCAPTCHA defends sites from bot attacks, and Google collects labeled images and uses it for training AIs in image recognition.
Although many processes are conducted by AI, it isn’t so self-contained as to give it control over something without a human-in-the-loop element.
AI is on the early stage of development. Andrew Ng said “There could be a race of killer robots in the far future, but I don’t work on not turning AI evil today for the same reason I don’t worry about the problem of overpopulation on the planet Mars”.
We all remember fantastic movies about AI failures without a human in the loop. Termainator, Star Wars, For instance, HAL9000 in the iconic movie 2001: A Space Odyssey could not do anything good without human.
HAL 9000: “I’m sorry Dave, I’m afraid I can’t do that”
In general, people are essential for supporting of AI. We will help AI and be friends with its. Nevertheless, nowadays even when AI is created with good intentions, trained to interact with people and to learn from them, it sometimes leads to bad results. Theodore, in the movie She, started to communicate with an AI bot called Samantha after divorcing his wife. After a long communication, Theodore fell in love with the AI bot, and Samantha loved him in return. However, nowadays we have problematic interactions with AI in the real life. The AI “Teen girl Tay” tweetbot, developed by Microsoft, is another demonstrative example of what AI can do without a well-intentioned human. Twitter users trained Tay to love Hitler, hate feminism and use obscene language within a single day. During the course of the day, Tay generated many racist and sexist tweets.
Self-driving cars are one of the most popular AI industries. Nevertheless, self-driving cars have limitations. They have learned to understand road signs, but small changes that vandals can make can lead to fatal consequences. Below you can see how researchers changed a stop sign in attempts to confuse the AI system.
In the first example, all AI systems recognized the stop sign as a “Speed Limit 45”. In the second and third cases, 67% of AI systems recognized the signs as “Speed Limit 45”. In the fourth example, AI systems were defeated by camouflage that was painted on the sign indicating a right turn. All AI systems recognized this sign as a “Stop”.
Despite its leading position in the self-driving car industry, even Tesla has had a fatal accident: The instructor of a self-driving beta-testing car, instead of following the road, was distracted by watching a movie, and the Tesla AI could not recognize danger and react in time to avoid the trailer of a truck.
In conclusion, we would like to say that AI cannot replace human intelligence and currently requires human control. However, use of AI is necessary, together with human skills like intuition and intelligence, to solve problems requiring speed, consistency and constant force.
Follow us on Twitter, Telegram, Facebook, Reddit and LinkedIn or visit our website to know more about DBrain and the future of Artificial Intelligence.
