It seems that everyone is talking about AlphaGo—the Google artificial intelligence (AI) system that recently defeated South Korean Go Master Lee Sedol. You’ve probably also heard of self-driving cars, artificial neural networks, and voice recognition. These technologies are all closely related to the development of artificial intelligence (AI). Even some museums are starting to use AI-enabled robots as virtual human guides to discuss the content of a museum exhibit and model human emotions. So, you keep hearing about AI, but what really is this technology all about?
What is AI?
According to John McCarthy from Stanford University, a leading AI scientist, “artificial intelligence is the science and engineering of making intelligent machines, especially intelligent computer programs.” In this definition, intelligence refers to “the computational ability to achieve goals in the world.”
AI is also about making guesses under great uncertainty. According to the managing director of Microsoft Research Eric Horvitz, one of the most important revolutions that accelerated recent research in AI was a shift away from focusing on programs based on the traditional “if a, then b; if b, then c” process, to instead focusing on programs built on probabilities.
There are many subsets of AI research, ranging from the speech recognition and natural language understanding (demonstrated by Siri and Cortana), to machine learning and deep learning essential to business analytics and systems designed to make sense of big data. A popular and relatively recent AI concept today is narrow AI, a subset focused on solving specific and well-defined problems, such as mastering Go as AlphaGo did. Narrow AI is an opposing concept of general AI that has full human intelligence and is more biologically inspired.
Cross Sections and Applications
The field of Robotics has always been closely related to AI. The first AI robot Shakey (1970) was developed by SRI and driven by a problem-solving program called STRIPS. Sony’s robotic dog AIBO (1999) was probably the most famous early example of AI. In more recent years, social robots have become more popular, including MIT’s sociable robot Kismet. This robot can “look for brightly colored objects, recognize the human face and express basic emotions through changes in its facial expression.”
In 2009, Ada and Grace, two responsive virtual humans, debuted at the Museum of Science in Boston. They were created by the University of Southern California’s Institute for Creative Technologies through computer-generated character animation, artificial intelligence and autonomous agent research. According to their developers, “not only are they capable of discussing the science content of a museum exhibit, they also can be funny and model a convincing range of human emotions, providing an unprecedented opportunity to inspire youth and learners of all ages about computer science and related STEM fields.”
Big data is another area in which AI has been making huge contributions. AI can help “sift through data and turn it into information that might otherwise be inaccessible,” thanks to the skyrocketing rate of data generation. Conversely, the confluence of big data and the social media explosion has made data storage easy and inexpensive, and fueled AI research.
As the 2014 TrendsWatch of the American Alliance of Museums pointed out, non-profit organizations have already been creating data sets that draw on museum data. The Tate and the Cooper Hewitt Museum have made their data accessible by “providing free access to datasets, applying Creative Commons licenses to digital content, or creating APIs (application programming interfaces) that allow programmers to build their own software on the museum's data.” Another project of this kind is from Bengler in cooperation with the Norwegian National Museum. This team applies machine learning processes and deep neural networks to museum collections to lay out maps of paintings based on their perceived similarity. According to the developer, the data is “promising and can be used to build novel user interfaces to aid in comprehension and exploration of collections.”
From artificial neural networks, to self-driving cars, and even museum humanoid robots, AI is a cutting edge technology at the intersection of medicine, robotics, linguistics and a multitude of other fields. MIT’s scientist Andrew McAfee (renowned as the coauthor of the bestseller The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies) even predicted on the radio show “Planet Money” that in 20-40 years vehicles would be driving themselves; mines, factories, and farms will be largely automated; and we will have a new economy that doesn’t require much labor. On the other hand, John McCarthy believes that “fundamental new ideas are required before AI can reach human-level intelligence, rather than just needing large databases and faster computers.”
Not everyone is optimistic about the future with AI. In 2015, Microsoft co-founder Bill Gates revealed his concern on the threat of artificial intelligence on human civilization; Stephen Hawking, Elon Musk, Steve Wozniak and others claimed at the International Joint Conference that artificial intelligence could “potentially be more dangerous than nuclear weapons.”
Will super-intelligent machines beat mankind in the future? More people are joining the discussion. However in the short run one thing seems certain: we can expect to see important developments and changes in our lives driven by artificial intelligence. In fact, in the last two years we have witnessed a race to set up AI research facilities. Big companies like Google, Microsoft and Apple are hiring more AI experts and focusing on machine learning to take control of the AI market.
While entrepreneurs and scientists are quite enthusiastic about artificial intelligence, museums, as one of the most important agents spreading knowledge and embracing science and technology, will almost definitely be influenced and should welcome the changes and challenges it brings.