The term "character AI" is not used extensively online – in fact, popular Google results for this term seem to come almost exclusively from a piece written in RoboHub – however, many IT professionals have an understanding of what character AI means, and some companies are choosing to make significant investments in this type of technology.
Character AI is predominantly defined as artificial intelligence work that builds characters with attributes and behaviors that humans see as authentic, whether they are ascribed to humans or independent robotic entities. In other words, character AI is the quest to build interactive characters with special intelligence, often through looking at human behavior and modeling it in some way.
One of the biggest uses of the term "character AI" is in gaming, and this is one area where companies might invest in building this kind of artificial intelligence.
In complex games, building character AI means establishing and implementing non-player characters within the game. These characters are not controlled by humans – they are autonomous and auto-generated by the game itself. However, as non-player characters, they share attributes with player characters – so that sometimes the human players can't even tell whether a given character is controlled by a human or not.
This goes back to the classic idea of a accomplishing a "Turing test," named after the famous mathematician Alan Turing of the mid-20th century – where a Turing test completion means that a technology has tricked human users into thinking that it is controlled by a human.
So how might companies invest in character AI, and why?
The short answer is that character AI can be immensely useful in all sorts of industries – in answering phones, playing games with humans, serving underlying robotics platforms, and in so many different ways that will help artificial intelligence interact with humans on a more deep and profound level.
Another answer for why companies might invest right now is related to building advanced "character" solutions with machine learning and artificial intelligence tools.
Take machine learning – the RoboHub piece mentioned above and other authoritative pieces on character AI show that one of the most popular ways to create character AI is to use machine learning and data provided by human-controlled technologies.
The process often works this way – the companies and stakeholders in question will build behavioral models by simply observing a human controlling technology – whether it's a robot or anything else. The engineers will take all of that training data and put it through machine learning algorithms, and the result will be character AI – artificial intelligence that acts like a human being.
In short, even though the term itself isn't popular, character AI is likely to enjoy popular usage in tomorrow's technology world. That's why so many companies may be considering refining it and putting it into consumer products and other business products and processes.