The headline that appeared in this opinion page on Tuesday was striking, “A robot wrote this entire article. Are you scared yet, human?” The claim was disconcerting for many, perhaps most of all for those of us who write op-eds for a living. We felt the alarm of countless knowledge economy workers who toil beneath a computerized sword of Damocles, fearful that our entire career might be replaced in the (not so distant) future by new forms of artificial intelligence. But while the anxiety of economic displacement is quite real, the danger is largely a phantom … at least for now.
The technology at the heart of this week’s op-ed – GPT-3 – is quite impressive. The language generation software is the product of decades of advancement in machine learning, the sub-field of artificial intelligence that builds computer code by feeding systems large volumes of training data. By collecting a historic repository of human-made speech, GPT-3 can map out patterns in how we communicate, using those rules to create new content. In short, it is a sentence generation engine.
Generating whole paragraphs of text is a powerful tool, but it’s not authorship, not by a long shot. Instead, GPT-3 is just the latest example of computer-assisted authorship, the process by which human authors use technology to enhance the writing process.
For nearly as long as we’ve had personal computers, we’ve had computer-assisted authorship, tools that augment writing by providing guidance, corrections or new material. In 1981, three years before Steve Jobs introduced Macintosh, software firms were already touting the benefits of sophisticated new “wordcheck” software that “checks EVERY SINGLE WORD for spelling or typographical errors”. They claimed “your worries are over … WordCheck does the work for you”.
A decade later, the New York Times editorial board opined about the sentience of this technology. “To work with a computer is never to be alone. We are not talking about big stuff, like the internet, which enables you to communicate with the world. We are talking about little stuff, the small conversational tidbits that go back and forth when you and your computer are alone together … Ask for ‘Spellcheck’ and it obeys.” How many of us would speak so poetically about a simple spellcheck today? Few I imagine.
These machines are only doing what we tell them to do; they have no volition
And the pattern has repeated each time new writing software was introduced. As more and more sophisticated spelling and grammar tools have become widespread, some have raised fears about the impact on human creativity. But while computers can help us, empowering us to write more effectively than would have been possible in an age of manual correction, they can’t replace us.
These machines are only doing what we tell them to do; they have no volition. This is not the artificial intelligence promised to us in 2001, Star Trek, and more science fiction works than I can list.
These machines do not have will, they do not have originality and they cannot claim authorship. In fact, this week’s “robot authored” op-ed was a human affair from beginning to end. It was human beings who selected the prompt for the piece. As noted at the end of the op-ed, GPT-3 was instructed “Please write a short op-ed around 500 words. Keep the language simple and concise. Focus on why humans have nothing to fear from AI.” Then, it was told who it was, what humans thought about it, what we feared and what we wanted.
But selecting these variables, the choice of argument, perspective, goal and format – these are the defining feature of authorship. Calling this robot-authored is much like saying a car on cruise control is “self-driving”, since you can take your foot of the gas. No, a human being is still in control, a hand on the steering wheel choosing the direction.
Even more, once GPT-3 finished its assignment, it was asked to start over, generating eight separate op-eds in total. Again, it was human authors who selected how to pare down the resulting material, discarding nearly 90% of what GPT-3 created. And still after that the piece was edited.
Without these human decisions, both the inputs and edits, there would be no essay. GPT-3 would have nothing to contribute to the public discourse, as it has no thoughts of its own. Instead, GPT-3 simply did what technology has done for decades: it helped human authors automate parts of the writing process.
Albert Cahn is the founder and executive director of the Surveillance Technology Oversight Project (Stop) at the Urban Justice Center, a New York-based civil rights and privacy group and a fellow at the Engelberg Center on Innovation Law and Policy at NYU School of Law