The rise of chatbots like ChatGPT has sparked a new wave of fears about robots taking our jobs. So far it looks like they’ve got a little way to go.
Google’s new AI chat bot, Bard, has already made a booboo.
But there’s a lot they can do well. It can write essays that pass anti plagiarism software. It can write well-structured summaries and emails, and some people even swear by song lyrics, recipes and even jokes. If that’s a glimpse of the future, a way in trouble, could AI chat bot really take people’s jobs?
Chat GPT and tools like it work with words and information. So it makes sense that when we think about the kind of jobs it will affect we’re mostly going to be thinking about information jobs. There might be other types of robots coming off the manual work. But the chat bots are coming after those of us in the laptop class.
When TDA asked its readers whether they were worried about ChatGPT coming for their job, that’s pretty much what they told us. About 90% of those who said they were worried were in information jobs. And it’s not just them. In 2013, a major Oxford University study looked at the kind of jobs that might be at risk from machine learning tools. At the very top, where routine information jobs like telemarketers, data entry workers and even sports umpires, but even some less routine jobs, had high risk like accountants, market researchers and real estate agents. But there’s another one that frequently appears on the list: journalism. Could we be on the chopping block?
Right now, if you ask ChatGPT to do journalism, it won’t do it. It doesn’t have access to any news after 2021. But Bing’s new ChatGPT powered AI assistant Sydney and Google’s new assistant, Bard, while not yet publicly accessible, are supposed to be more up to date.
So to put chatbots to the test we asked ChatGPT to summarise an event from 2021: The fall of Kabul, and we compared it to our own coverage. The limitations of the chat bot are clearly on show here. It can collect information, write it up clearly and fill in any blanks to produce something that looks and feels like a news story. But at least so far, it isn’t able to bring a unique perspective or provide nuanced analysis to any write up of a news story.
A lot of journalism skills, like speaking to sources, evaluating information, investigating stories, and putting it all in context – chatboxes can’t do that. They also can’t produce any original stories. The strengths and weaknesses point us to a different way to think about the potential of chatbots, rather than simply focusing on whether it will take over our jobs or not. We can ask how it can complement the sorts of things humans are good at.
For example, if chatbots could do some of the more boring and mundane tasks that we would rather not do, it could free us up to do some of the more creative aspects of our jobs. As three Harvard researchers put it. The right question might not be what jobs a robot will do, but what jobs robots can help humans do. There is likely to be hardship and economic pain for those who struggle to adapt. But over the long term, they could unlock jobs we haven’t even thought of yet.
So working life with ChatGPT could have pros and cons, but it’s worth finishing with a note of warning. While it’s certainly useful to think about how today’s technologies will shape the future, we should remember our crystal ball gazing is often because the future is shaped as much by the things we don’t see coming as the things we do.
Warnings of robots taking jobs have been around for centuries, as have dreams of robots delivering us to some perfect utopia where all our needs are taken care of. Neither has turned out. Take that Oxford study that looked at which jobs might be most at risk. That was in 2013 and it predicted nearly half of jobs would be obsolete by 2033. We’re halfway to 2033 and that prediction hasn’t aged so well. What may be more important than trying to predict the future is being ready to adapt to it.