Review | Creating Human-Like AI with 87% Accuracy

Computer scientists are nonstop working on improving artificial intelligence software (AI) to ease the lives of humans.

Modern AI is found on phones and computers, most either have Siri or Google, that answers simple commands and questions. To further improve humans’ lives, scientists are trying to find ways to teach AI the ways of humans nonverbal cues, currently with a 70 to 87% accuracy, according to Maria Temming in Science News. At the moment, AI cannot discern subtext, sarcasm, or other nonverbal cues within our tones and actions and this is a hindrance to AI’s capability. Without understanding nonverbals, AI persons will not be able to have an honest human conversation with another human.

This may mean that AI may not be able to create its own human-like nuances, such as sarcasm or tone of voice. Or, that with which, AI will soon be able to speak as a human would speak. Similarly to computer-mediated communication, expressing nonverbal cues is integral to true communication and because of this, it is necessary for computers to learn this form of language.

Artificial intelligence software is a controversial subject due to its relation to creating life. Scientists, who write incredibly complicated software, essentially produce a form of life that is capable of thinking for itself. What they want to do is teach them more on how to be more like humans in order to better understand humans. Understanding nonverbal communication, like paralanguage, will help simple things like talk-to-text or asking Siri questions as well as complex things like manufacturing a non-carbon life form that eerily resembles a human.