According to Joseph Wilson in his article “Why AI Will Never Fully Capture Human Language,” researchers in artificial intelligence have made amazing achievements in emulating human language. However, they still need help to capture the parts that genuinely comprise human language. He provided several justifications for his essay title. Wilson’s first argument was that anything must be written down for a computer to perceive it as a language. Software programs called neural networks, which are intended to identify patterns in huge data sets, are programmed into computers that are capable of interacting with humans or creating what may be considered to beat poetry.
Neural networks develop the ability to repeat the patterns they discover over time. For instance, Goodwin “trained” the AI that created the road trip novel using a library of 60 million word novels and poetry. However, the training ignores all non-verbal modes of communication, including body language, oral histories, tone of voice, sign language, and the larger cultural context in which individuals are conversing. In other words, it omits a lot of interesting information that makes nuanced communication between people possible.
Furthermore, Wilson said that writing is not universal. For example, only around half of the 7,100 natural languages spoken across the globe are written down. Audio recordings and speech recognition programs may help replace part of the gap, but in order for them to operate, algorithms must be trained on massive amounts of data, preferably from millions of different speakers. Oral languages are often derived from tiny communities that have traditionally been socially and geographically isolated. However, when these languages are added to databases, they must be transcribed and codified in textual form. The issue is that written words are never an exact depiction of how a language is spoken. When a language is initially transcribed, it is vital to select what dialect should be regarded as “standard” and to classify the many non-linguistic indicators that accompany spoken language.
The essay also tackles the fact that novel-writing cars and chat-bots meant for “natural language processing” have no understanding of languages whatsoever. Instead, they accomplish a narrow subset of linguistic competence, a reality that is sometimes overlooked when the technology media concentrates on sensational allegations of artificial intelligence awareness. Conversations in everyday life include the usage of a wide variety of communication signals. Real discussions are disorganized, with individuals talking over one another, vying for the opportunity to speak, and hesitating to find the appropriate phrase; they develop in a manner comparable to an unplanned dance.
Robots may never be able to write literature in the same manner that humans have throughout history. This is due to the fact that coming up with ideas for writing takes a high degree of creativity. The human brain spontaneously absorbs knowledge throughout our lifetimes to construct sophisticated neural networks. This balanced learning allows us to be creative by abstractly transferring information obtained in one setting to entirely another. This form of learning cannot yet be replicated in robots. For years, developers have attempted to define elements like culture and feeling in order to drive innovation, but the results have always been disappointing.
Machines can write an essay or design a picture, but they are rarely creative. They are continually following human instructions or working toward certain objectives. Instead of creating anything new, they employ aspects that already exist in their training data. “Why AI Will Never Fully Capture Human Language,” by Joseph Wilson, discusses the reasons why AI will never fully capture human language; nevertheless, artificial intelligence has taken such great advances in recent years that it is believed to generate a way for robots to write great literature in the future.