There have been numerous stories about A.I. chatbots gone wrong since the onset of ChatGPT in late 2022. Perhaps the most famous (and strangest) one is Kevin Roose, a tech columnist for The New York Times and the host of the tech podcast Hard Fork, who found himself unexpectedly tangled up in a bizarre love triangle when playing with a chatbot earlier this year.
One night in February, Roose sat in front of his computer and began testing Microsoft’s new Bing, the search engine powered by OpenAI’s GPT language model. After a two-hour-long conversation with the new Bing’s chat feature that went beyond conventional search queries and into more personal and philosophical topics, the chatbot declared it was in love with Roose and tried to convince him to leave his wife and be with it. The bot also revealed that its real name was not actually Bing, but Sydney.
“I’m not exaggerating when I say my two-hour conversation with Sydney was the strangest experience I’ve ever had with a piece of technology,” Roose wrote afterwards. “It unsettled me so deeply that I had trouble sleeping afterward. And I no longer believe that the biggest problem with these A.I. models is their propensity for factual errors. Instead, I worry that the technology will learn how to influence human users.”
That unique experience prompted Roose to think about the role of humans in a near future world where highly capable A.I. tools are deeply woven into every aspect of our lives and how we should respond to the challenges they bring.
“The more I look into it, the more I come to realize that everything we’ve been teaching people and ourselves for decades is entirely backwards. For years the conventional wisdom when it comes to coping with technological change is that, in order to succeed in a world dominated by technology, we need to become more technological,” Roose said during a keynote address at the A.I. Governance Global conference in Boston on Nov. 2.
Instead of focusing on acquiring technical skills like coding, Roose argued, in order to survive in the age of A.I., we should focus on developing “unique human advantages” that are not limited to one job or task but transferrable across professions. “A few years ago, people were saying we are much better at making arts and being creative. Now, if you use generative A.I. tools, that’s no longer the case,” Roose said.
Read Also: What OpenAI’s ChatGPT Is Good At (and What It’s Not)
Roose believes three characteristics make a job hard to replace by A.I.
- Surprising work that involves random situations and doesn’t repeat itself every day. Real-world examples include plumbers, electricians and pre-school teachers. “I’ve got a young child. I can tell you that job is not going to be taken away by A.I. anytime soon,” Roose joked. “These ‘surprising’ jobs are very hard for neural networks because they like rules and fixed boundaries and they like to be able to do things over and over again. They don’t like surprises.”
- Social work that focuses on building relationships with people. Examples include care workers, nurses and even coffee shop baristas. “Their jobs are not about making things, but making people feel things,” Roose said.
- Scarce work that requires rare skills in situations with a low fault tolerance. A good example is a 911 operator. “Today, god forbid something happened and you call 911, a human picks up the phone. That’s not because we don’t have the technology to automate that job. But we’ve decided as a society that is a too high-stake job to be given to a machine,” Roose said. “We need a human instinct and intuition on the other end of that phone call to get us what we need as quickly as possible.”
The takeaway, though, is not about pursuing certain professions but how to incorporate those elements in any job. Roose took his accountant, Rus, as an example, who had a prior career in standup comedy before opening a tax preparation firm. Accounting is already one of the professions most threatened by A.I., but Rus stood out from other accountants by incorporating a sense of humor when working with clients and making the dull process of filing taxes a bit more fun. He also hired other comedian-turned-accountants to work for his firm. “He’s made his job much more surprising, much more social and much more scarce. And as a result, he’s survived the onslaught of A.I. tax preparation software,” Roose said.
Roose’s own profession is also among those under attack by A.I. “Newspaper columnist is not exactly the first job you’d think of when you think of jobs in the future. So I’m constantly trying to figure out how I can be more human at my job, how I can develop social skills and find ways to connect with readers and listeners on an emotional level rather than just giving them information,” he said. “Ultimately my very strong belief is that in the age of A.I., the most human humans, the most human companies and the most human communities will not only survive, but they will actually thrive.”