2. Narrow, General, and Super AI (Dana) - Lucia922/IT115-WiKi GitHub Wiki
AI: Definitions and General Terms
Defined broadly, Artificial Intelligence (AI) is the simulation of human intelligence and problem solving ability in computers and computer controlled machinery. AI can be classified further into several categories: Artificial Narrow Intelligence, Artificial General Intelligence (AGI), and Artificial Super Intelligence (ASI). The last two categories are as of yet still theoretical. Despite the predictions of many early AI pioneers like Marvin Minsky that human level AI (AGI) was just a few years away, it’s proven difficult to achieve. In fact, how the phenomenon of intelligence arises in biological organisms has proven to be an incredibly complex problem. We still aren’t sure exactly how it happens. Artificial Super Intelligence (ASI), as seen in science fiction like Sky Net from the Terminator movies or speculation by futurists such as Ray Kurzweil, seems even further beyond the current technical capability of humans.
The Rise of Narrow AI
What has seen major attention in the past few years are advances in the field of Artificial Narrow AI. This is AI that is programmed to perform a narrow set of functions really well, hence the term. Narrow AI doesn’t approach levels of human general intelligence, but does have the capability to learn how to better solve the range of problems for which it’s designed. This is in contrast to humans and General AI, which would be able to apply learning and problem solving ability to any topic.
Recent investment in Narrow AI has been driven by business and industry, with the hopes of paying off in terms of increased shareholder value, an enhanced customer service experience, and reduction of operating expenses. It is used in applications such as chatbots, identifying patterns in large amounts of data, and character recognition. The machine driven pattern recognition capabilities of narrow AI have also been used to help doctors identify possible tumors that they might otherwise miss. Narrow AI has seen widespread use in augmenting the capabilities of skilled workers, while also automating routine tasks that can now be done automatically. Some examples of this automation would be in the fields of data entry and customer service.
Narrow AI Methods
Machine Learning: This is where an algorithm parses through immense amounts of data and the algorithm identifies patterns and makes predictions based on the data that’s fed into the algorithm. The AI learns and adapts to new information in a process called unsupervised learning, where human oversight isn’t required. Machine learning is commonly used to predict the behavior of financial markets.
Natural Language Processing: AI that allows computers to interpret human language. Commonly used in chatbots on websites and automated phone support.