How AI and the Lure of Efficiency Threaten Human Writing

By transforming the way people write, artificial intelligence is changing what it means to be human.

Machines have been imitating human writing since at least 1953, when Christopher Strachey, an English computer scientist programmed a Ferranti Mark 1 computer to pen love letters. Those missives weren’t particularly convincing, but today, advances in AI mean technology can competently produce a variety of written materials that used to require human intelligence: marketing copy, real estate listings, quarterly earnings reports, sports reporting, legal pleadings, and more. Writing has always been the realm of humans?— until the rise of AI. As AI becomes increasingly capable of using human language, it’s altering perceptions of what it means to be human.

“It’s important not to let fascination with AI glitz or awe in the face of its efficiency push us to abandon what we value about human writing.”

The widespread use of AI to perform language tasks that humans used to do also raises questions about the effects on humans themselves. When people fail to make use of their own writing capacities, they pay a price. Neuroscientists have found that writing changes brain function. For example, the caudate nucleus —?an area of the brain responsible for higher-level?learning, planning, and memory — is more active in professional writers’ brains than in novice writers. Scientists don’t yet fully understand the impact that a decline in human writing might have on thinking?and human intelligence, but society shouldn’t be quick to?dismiss it.

AI tools are already disrupting writing professions, and the ultimate impacts are impossible to predict.

Research shows readers can’t always tell whether human journalists or algorithms produced?news stories?— though?they often rate the algorithmically generated writing as more boring. Professional journalists have voiced concerns regarding the quality of?AI writing, noting that AI systems frequently rely on single sources of data, lack creativity, and fail to capture the nuances of people’s words.

Nonetheless, AI is making inroads into the writing profession. The US Department of Labor estimates that American writers — ad copywriters, grant writers, speech writers, and more — collectively earn more than $675 billion every year. As AI permeates writing tasks, it?disrupts?the professions that focus on writing, placing these wordsmiths at risk. Some media professionals have already lost jobs to AI. For example, in 2020, Microsoft replaced roughly?50 human contract news producers with an automated system that assumed the job of selecting news stories for the MSN website and determining their placement.

“One of the biggest dangers from today’s AI language prowess is that we forget what applying our writing skills can do for our intelligence and psyche.”

AI is also affecting the?legal profession, as law clerks and associates — typically responsible for discovery, argument construction, legal research, and writing — can now offload many of their tasks to AI. New tools can perform work that’s practically impossible for human lawyers and paralegals to execute: LexisNexis’s Context uses natural language processing to create arguments targeted to appeal to specific judges’ ways of thinking, and Attorney Analytics?calls on a gigantic database to tell lawyers how other attorneys have argued cases similar to theirs.?

The jury is out?on AI’s?impact on writing professions. Some professionals see the value in AI taking over menial tasks so they can dedicate themselves to more complex ones. Others fear their own obsolescence.

AI won’t replace human writing — or even translation — completely.

People write for myriad reasons — money, love, self-expression, and more. As a tool of everyday life, writing serves to help people remember to-dos, connect with friends, or track activities in a journal, for instance. Students write practically daily to fulfill assignments. Other people, such as authors and reporters, write because they’re paid to do so or in order to advance in their careers.

People also write for heartfelt reasons: to share information or knowledge, to release or express emotions or ideas, to make sense of the world, to make sense of oneself, or simply for enjoyment. People will continue to write because their motivation often involves more than merely producing a piece of writing — they have reasons to engage in the process itself.

“I write entirely to find out what I’m thinking, what I’m looking at, what I see, and what it means. What I want and what I fear.” (American essayist Joan Didion)

The limits of AI translation point to the uniquely human ways people create and use language. Digital translators have made great strides in recent years, Nevertheless, they still often make mistakes in grammatical gender and have trouble interpreting the nuances of words used in different contexts.?Sometimes, no adequate way exists to translate words from one language to another, because the concepts behind the words are culturally distinct. Learning a foreign language provides insight into culture and history, and it can open a person’s eyes to alternative perspectives. As digital translation has become more popular, the interest of American college students in learning languages has plummeted — an unfortunate development, as studying foreign languages offers profound benefits for learners and society.

The human brain inspired the technology behind AI.

The technology behind AI was inspired by the human brain. The term “neural network” refers to?a programming model that imitates the structure of connections?between neurons in the human brain. When a neural network has more than one layer, it’s referred to as a “deep” neural network. “Deep learning” denotes machine learning that uses deep neural networks.? “Natural language processing” (NLP) refers to the use of computers to generate human language and to simulate the understanding?of it. NLP depends on a tool called a transformer — an algorithm trained on millions of data points to perform certain functions, such as translation or answering questions.

“The invention of writing was a big deal…Being literate conferred a magic light sword for thought, exchange, and action. Now AI wields its own light sword.”

NLP enables AI systems to accomplish many linguistic tasks, including the following:

• Speech — AI systems can render text as spoken words — for example, to narrate an audiobook — or?transcribe spoken words as text. AI systems can also mimic individual humans’ vocal patterns, based on recordings — a technology that one company is using to enable people to “speak” with their departed loved ones.
• Writing — Transformers can anticipate what a user might type next,?or they can generate entirely new text (for example, limerick generators). Additionally, they can be programmed to churn out?article summaries and?news stories.
• Search — AI transformers can be trained to use pattern recognition to predict what human users want to know — the technology underlying Google Search.
• Translation — NLP also enables AI systems to translate text or spoken words from one language to another.

AI can mimic human creativity but can’t replace it.

To gain insight into whether AI threatens to replace humans who work in creative professions, it’s necessary to define creativity. A common formula holds that creativity results in a product that’s new and that has value. In the 1990s, psychologists began differentiating between “Big C” and “little c” creativity: Big C creativity occurs in areas such as music, literature, and the arts, spurring innovation on a broad societal level. Little c creativity, on the other hand, includes innovative acts on a more?personal level — for example, improving a favorite recipe. Cognitive scientist Margaret Boden expands the definition of creativity by distinguishing between “historical creativity,” resulting in something new in human history, and “psychological creativity,” which results in something new in your own thinking.

“Writing makes our words last. It captures things we say but also embodies its own character and style.”

AI has already demonstrated its historical creative capabilities. For example, the AI system AlphaGo beat a human champion at Go, a complex and ancient strategy game, by using creative, unexpected moves, and AlphaFold triggered a scientific breakthrough in the understanding of protein folding. But AI has?yet to create?art that has reshaped culture in a?Big C creative way. Many people believe?creative works should capture the authenticity of their maker, which cannot be simulated; authenticity, they hold, comes from creativity as a process arising from life and experience.?Defenders of human creativity want artworks that capture the emotion and intensity associated with the human creative process.

The use of AI in education is illuminating questions about authorship and AI’s social impacts.

As AI tools serve as both author and editor of written work, the meaning of authorship is beginning to shift. In universities, students’ use of AI to complete assignments is spurring administrators to create policies that define what it means to write a paper. However, it is difficult to determine whether a student or an AI system has written a text. The ease with which students can now cheat with impunity is prompting concerns for the future of education. “This technology can clearly undermine the integrity of all education,” said Morten Irgens, a Norwegian AI expert and university administrator.

“AI is incapable of pausing, thinking, and rewriting, except perhaps in the sense of tinkering with sentences humans have written.”

After the launch of ChatGPT, members of the Association for Norwegian Language Arts Education prompted a national debate by writing a letter to Norway’s parliament expressing a fear that AI text-generation tools could threaten Norwegians’ literacy skills, intellectual life, and democracy. According to these critics, writing serves a critical role in students’ learning process, prompting students to engage actively — rather than passively — with information and ideas. To circumvent the threat AI tools pose to learning outcomes, they recommended giving assignments that require students to assemble and synthesize information themselves.

Human-centered AI keeps people at the helm, making AI a tool to expand humans’ capabilities.

One way to think of AI’s potential is simply as a tool that expands the user’s own abilities, extending the person’s cognitive capacity and opening up opportunities for “human-AI co-creation.”?Mira Murati, OpenAI’s senior vice president of research and product, highlights the potential for AI to free up professionals to create and innovate by taking over their menial work. The concept of “human-centered AI” is emerging — AI that promises to improve people’s lives rather than threatening to replace them. AI research initiatives such as Stanford’s Institute for Human-Centered Artificial Intelligence?and UC Berkeley’s Center for Human-Compatible Artificial Intelligence focus on the role AI could play in improving the lives of humans.?

“I never understand anything until I have written about it.” (18th-century English writer Horace Walpole)

The AI tool Sudowrite provides an example of the ways humans and AI systems can collaborate and co-create. Sudowrite is advertised as a “brainstorm buddy,” designed to help users overcome writer’s block by generating ideas that stimulate creativity. Another AI tool, Marlowe, helps writers edit their novels, giving feedback on pacing, the personality traits of characters, and more. As the designers of these tools, humans maintain a clear advantage: They get to choose the roles both humans and machines play in this new age of human-AI interaction.

Adopting “rules of the road” for the use of AI in writing could help humanity balance the technology’s benefits and risks.

Determining how to balance the potential benefits of AI technologies with the new kinds of risks and challenges they pose is complex, and there’s no?road map to navigate this new terrain. Frank Pasquale, a Cornell law professor, has suggested adopting laws for AI and robotics, such as “Robotic systems and AI should complement professions, not replace them” and “Robotic systems and AI should not counterfeit humanity.” The “BOT bill,” legislation adopted in California in 2019, requires disclosure of the use of automated programs to influence people’s choices when voting or shopping. Similar disclosure requirements for writing could protect the written word from AI counterfeits.

“Human writing is a light sword for our minds and for connecting with fellow humans. It’s on us to retain that sword’s brightness.”

When you write, you shouldn’t feel pressured into using AI technology if it doesn’t serve you. Writing is about more than merely communicating: It’s a way of processing your thoughts and being in the world. Don’t cede control to AI or let it take over your voice. When it comes to your writing, you have the last word.