Artificial Intelligence

Introduction
Artificial Intelligence is the simulation of human intelligence by machines. The technology is advancing rapidly, and the gap between machine and human mind may soon be nonexistent. It can be difficult to predict what trajectory the technology will take, but it’s certain to transform the field of Technical Communication for better or worse. There are ample opportunities for AI to be used to assist the work of a Technical Writer, but there is also a prevalent fear that AI may soon become proficient enough to make human writing obsolete. Whether AI is perceived as a friend or foe, it will be pertinent for all technical communicators to be familiar with what AI can do, and which skills still remain exclusive to the human writer.

Current Uses
Many modern uses of AI are examples of machine learning (ML), and the distinction between ML and AI is often poorly understood. While AI aims to simulate a human writer, ML uses algorithms to learn from data without being explicitly programmed. AI employs knowledge-based learning, in which the human teaches the machine what it knows, whereas ML can figure out how to accomplish a task on its own when trained with the proper data. Until fairly recently ML was a theoretical concept, without enough data available to test its full potential. In recent years it has become widespread, but concerns remain that ML techniques fail to account for rhetorical context and need ethical principles to be reinforced by a human agent or better incorporated into the technology.

Predictive Text
Predictive Text is a prevalent example of ML, in which an algorithm makes an educated guess at what the next word the writer types will be. Though this can speed up the writing process, it’s also been found to encourage more predictable writing. There are cases in which the suggestion overrules the writer’s original word choice based only on its convenience, and cases when the suggestion is rejected at the cost of writing flow being interrupted. Still, if one is employing plain writing principles or aiming for quickness and accuracy, it can be a highly useful tool.

Spelling & Grammar Check
Writing assistants like Grammarly & Smart Compose rely heavily on ML to improve the efficiency of their programs. Though some writers believe a rigid adherence to grammar can interfere with their unique voice, this isn’t a concern for ESL writers, Grammarly's most frequent users. Google’s Smart Compose aims to solve the issue of a grammatically correct but impersonal writing voice by analyzing the writing of its user and simulating their specific style.

Chatbots
Chatbots are a prime example of AI, as this technology seeks to directly imitate a human writer. Nowadays, conversational agents come in many forms, some speech-based like Siri or Alexa and others text-based customer servers. It’s been found that the more human a chatbot sounds, the more positive the response of the human at the other end. This has led many companies to focus on infusing AI with humor and conversational quirks in order to better replicate the experience of interacting with a human.

What the Future May Hold
Many of the limitations of AI may turn into strengths before long, but for the time being humans still outstrip AI in many categories. AI still fails to write with the nuance of a human and cannot compete in subjective or persuasive writing. AI cannot yet create audio or video, two of the more common ways people consume digital media nowadays. There are impressive code-writing AI programs, but thus far they are still as prone to error as human coders. Still, it isn't a matter of if AI will surpass the abilities of a technical writer, but when.

Collaborative Writing
Though many take an adversarial approach to AI, others are embracing the possibility of cooperation between humans and machines. AI is already capable of autonomously performing simpler tasks and could be seen as an ideal collaborator for humans when working on more complicated projects. While many science journalists are concerned by the number of research articles being written by bots, Narrative Science is a company that’s utilizing AI to make sense of patterns in the data, a tedious job formerly performed by humans over much longer periods. By deploying ML techniques an agent can anticipate the needs of its collaborator without being delegated to and can filter data by relevance with much greater efficiency than a human. This process of co-writing may soon become the norm.

AI in Development
OpenAI is currently the most prominent company in the development of AI writing. GPT-2 is their latest machine learning writing assistant, and while it’s highly powerful it contains only one and a half billion neurons compared to the hundred billion of the human brain. OpenAI claims that GPT-3 is complete and will close the gap between AI and the human brain, but they are waiting to release it until they can ensure that the program’s benefits be “distributed as widely as possible".

It is estimated that in five years' time AI will be capable of writing a high school level essay, and in twenty years' time may be proficient enough to write a New York Times bestseller. Technical communicators must adapt to working alongside machines, or designing interactive programs to be better at simulating human intelligence. Being prepared for the changes and attuned to the possible benefits and pitfalls of AI is essential for anyone entering the field of TC.