In December, the staff of the American Writers and Artists Institute—a 26-year-old membership organization for copywriters—realized that something big was happening.
The latest edition of ChatGPT, a “big language model” that uses the Internet to answer questions and perform tasks on command, had just been released. His abilities were astonishing – and squarely in the bailiwick of people who generate content, such as ad copy and blog posts, for a living.
“They’re shocked,” said Rebecca Matter, the institute’s president. During the holidays, she rushed to host a webinar on the pitfalls and possibilities of the new artificial intelligence technology. More than 3,000 people applied, she said, and the overall message was cautionary but reassuring: Writers could use ChatGPT to complete assignments faster and progress to higher-level roles in content planning and search engine optimization.
“I really think it will minimize short copy projects,” said Ms. Matter. “But on the other hand, I think there will be more opportunities for things like strategy.”
OpenAI’s ChatGPT is the latest advancement in a steady march of innovations that have held the potential to transform many professions and wipe out others, sometimes simultaneously. It is too early to add up the opportunities and threats, or measure the overall impact on labor demand and productivity. But it seems clear that artificial intelligence will impact work in different ways than previous waves of technology.
The positive view of tools like ChatGPT is that they can complement human labor rather than a replacement. However, not all employees are optimistic about the future impact.
Katie Brown is a suburban Chicago grant writer for a small nonprofit focused on tackling domestic violence. She was shocked to learn in early February that a professional association for grant providers was promoting the use of artificial intelligence software that automatically completes parts of an application, requiring a person to simply polish it up before submitting.
Called Grantable, the platform is based on the same technology as ChatGPT and markets itself to freelancers who charge through the application. That, she thought, clearly threatens opportunities in the industry.
“For me, it’s common sense: Which one do you think a small nonprofit will choose?” said Mrs. Brown. “A person with a full-time salary plus benefits, or someone equipped with AI that you don’t have to pay benefits for?”
Artificial intelligence and machine learning have been working in the background of many companies for years, for example to evaluate large numbers of possible decisions and to better match supply and demand. And numerous technological advances over the centuries have reduced the need for certain workers – though each time the jobs created more than compensated for the numbers lost.
However, ChatGPT is the first to confront such a wide range of white collar workers so directly and so accessible that people could use it in their own work. And it’s improving fast, with a new edition released this month. According to a survey conducted by job search website ZipRecruiter after the release of ChatGPT, 62 percent of job seekers said they were concerned that artificial intelligence could derail their careers.
“ChatGPT is the one that has made it more visible,” said Michael Chui, a partner at the McKinsey Global Institute who studies the effects of automation. “So I think it started to raise questions about where timelines could be accelerated.”
A new generation of chatbots
A brave new world. A new breed of chatbots powered by artificial intelligence has sparked a battle to determine whether the technology can turn the internet’s economics upside down, turning current powerhouses into has-beens and creating the industry’s next giants. These are the bots you should know about:
That is also the conclusion of a White House report on the implications of AI technology, including ChatGPT. “The main risk AI poses to the workforce is the general disruption it is likely to cause for workers, whether they find their work has been re-automated or their job design has fundamentally changed,” the authors wrote.
For now, Guillermo Rubio has noticed that his job as a copywriter has changed significantly since he started using ChatGPT to generate blog post ideas, write first drafts of newsletters, create hundreds of small variations of ad copy, and spark research on a topic about that he could write a white paper on.
Since he still charges his clients the same rates, the tool has simply allowed him to work less. However, if the usual rate of copy drops – which it may, as technology improves – he is confident that he can move into content strategy consulting in addition to production.
“I think people are more reserved and fearful, and there’s a good reason for that,” said Mr. Rubio, who lives in Orange County, California. “You can view it negatively, but you can also embrace it. I think the biggest takeaway is that you have to be flexible. You have to be open to embracing it.”
After decades of study, researchers understand a lot about the impact of automation on the workforce. Economists, including Daron Acemoglu of the Massachusetts Institute of Technology, have found that technology has played a primary role in increasing income inequality since the 1980s. As unions languished and training and retraining systems eroded, workers without college degrees saw their bargaining power diminish in the face of machines capable of rudimentary tasks.
However, the arrival of ChatGPT three months ago has sparked a wave of research based on the idea that this is not your average robot.
A team of researchers conducted an analysis of the industries and professions most exposed to artificial intelligence, based on a model adapted for generative language tools. At the top of the list were humanities professors, legal service providers, insurance agents and telemarketers. However, mere exposure does not determine whether the technology is likely to replace workers or merely enhance their skills.
Shakked Noy and Whitney Zhang, PhD students at MIT, conducted a randomized controlled study of experienced professionals in fields such as human relations and marketing. The participants were given tasks that usually take 20 to 30 minutes to complete, such as writing press releases and short reports. Those who used ChatGPT completed assignments an average of 37 percent faster than those who didn’t – a significant increase in productivity. They also reported a 20 percent increase in job satisfaction.
A third study – using a program developed by Microsoft-owned GitHub – evaluated the impact of generative AI specifically on software developers. In a trial conducted by GitHub researchers, developers were given an entry-level task and encouraged to use the program, called Copilot. They completed their task 55 percent faster than those who did it manually.
Those productivity gains are unlike almost any seen since the widespread adoption of the personal computer.
“It seems to be doing something fundamentally different,” said David Autor, another MIT economist, who advises Ms. Zhang and Mr. Noy. “Computers used to be powerful, but they simply and robotically did what humans programmed them to do.” Generative artificial intelligence, on the other hand, is “adaptive, it learns and is able to solve problems flexibly”.
Peter Dolkens, software developer at a company that mainly produces online tools for the sports industry, notices this very well. He integrates ChatGPT into his work for tasks such as summarizing bits of code to help colleagues pick up the project after him, and proposing solutions to problems that have left him speechless. If the answer isn’t perfect, it asks ChatGPT to refine it or try something else.
“It’s the equivalent of a very well-read intern,” said Mr. Daggers, who is in London. “They may not have the experience to know how to apply it, but they know all the words, they’ve read all the books, and they’re able to get some of it.”
There is another conclusion from the first study: ChatGPT and Copilot highlighted the least experienced employees the most. If true, it could mitigate the inequality-widening effects of artificial intelligence more generally.
On the other hand, as each worker becomes more productive, fewer workers are required to complete a set of tasks. Whether that leads to fewer jobs in certain industries depends on the demand for the service provided and the jobs that can be created helping to manage and direct the AI. Prompt engineering, for example, is already a skill that those who play around with ChatGPT can add to their resumes long enough.
Given that the demand for software code seems insatiable and developer salaries are extremely high, increasing productivity seems unlikely to deprive people of opportunities to enter the field.
That won’t be the same for every profession, though, and Dominic Russo is pretty sure it won’t be true for his own: writing appeals to pharmacy benefit administrators and insurance companies when they reject prescriptions for expensive drugs. He has been doing the job for about seven years now and has built up expertise with only on-the-job training, after studying journalism at university.
After ChatGPT came out, he asked it to write an appeal on behalf of someone with psoriasis who wanted the expensive drug Otezla. The result was good enough to require only a few edits before submission.
“If you knew what to prod the AI with, anyone could do the job,” Russo said. “That’s what really scares me. Why would a pharmacy pay me $70,000 a year when they can license the technology and pay people $12 an hour to run it in prompts?”
To protect himself against that possible future, Mr. Russo has built his side business: selling pizza from his home in southern New Jersey, an endeavor he believes will not be disrupted by artificial intelligence.
Yet.
Audio produced by Kate Winslett.
