For the last few weeks, chatbots powered by artificial intelligence have made waves on social media as users talk with these automated conversation partners. The bot ChatGPT, for instance, can provide reasonable answers to questions posed by users regardless of their spelling or grammar. These responses are delivered in natural-sounding sentences that do not require scripting, a significant improvement on clunky chatbots of the past. As a result, many big companies are looking to get in on the ground floor of this potentially game-changing technology. In fact, last month Microsoft announced that it would make a multibillion-dollar investment in OpenAI, the startup that launched ChatGPT.
The hope is that resources like these will someday be able to perform jobs usually done by humans, such as answering customer service questions. Still, experts suggest that firms should proceed with caution before placing too much trust in AI chatbots. While most scripted bots are trained to say some version of “I don’t know” to requests they cannot process, ChatGPT is more likely to deliver a response with complete confidence regardless of whether or not it is correct. “We don’t want to be in the bad answer business,” said John Willcutts of the customer-experience software company Nice Ltd. “A really bad answer in a very critical situation would be a very real problem.”
Even the chief executive of OpenAI has advised against using ChatGPT “for anything important right now.” According to a tweet from CEO Sam Altman last month, OpenAI’s bot is great for “fun creative inspiration,” but when it comes to “reliance for factual queries; not such a good idea.” Besides potentially delivering incorrect information, studies show that bots like ChatGPT are not very helpful in sensitive situations. For example, a chat service called Koko claimed it would provide users with emotional support through its app, only to face loads of criticism when it was later revealed that the startup relied on AI for its responses. “Once people learned the messages were co-created by a machine, it didn’t work,” said the co-founder of Koko in a statement announcing that the firm removed AI from its system. “Simulated empathy feels weird, empty.”
- What are the potential advantages and disadvantages of using AI chatbots in business settings like customer service?
- Do you think AI chatbots will ever be sophisticated enough to provide emotional support for human beings? Why or why not?