Can the new AI tool ChatGPT replace human work? Judge for yourself

A new artificial intelligence tool using natural language processing has captured the public's imagination, amassing more than a million users.

New artificial intelligence tool can respond to a human question better than predecessors, say observers

ChatGPT software highlights advances, limitations of modern artificial intelligence

10 months ago
Duration 1:59
ChatGPT is artificial intelligence chatbot software capable of writing poems, college-level essays and even computer code. Experts say the software highlights how far AI has come in just a few years, while still spotlighting concerns around accuracy.

There's a new artificial intelligence tool in town, and it's getting massive mainstream attention.

ChatGPT is a program where users can type in a question or a task, and the software will come up with a response designed to mimic a human. It was trained using billions of examples of text across the Internet.

"One of the key features that sets it apart is its ability to understand and generate natural language. This means that it can provide responses that sound natural and conversational, making it a valuable tool for a wide range of applications."

Or, so says the chatbot about itself — ChatGPT wrote the paragraph above.

How well the processing tool actually "understands" language is not clear. But it is turning heads.

"You can have what seems alarmingly close to a human conversation with it, so I was a little taken aback," said Osh Momoh, chief technical advisor for MaRS, an innovation hub in Toronto.

A person uses ChatGPT on their laptop.
A new artificial intelligence tool called ChatGPT, released Nov. 30 by San Francisco-based OpenAI, allows users to ask questions and assign tasks. (CBC)

The tool was created by OpenAI, a San Francisco-based research and development firm co-founded by Elon Musk that counts Peter Thiel and Microsoft among its investors. 

ChatGPT has captured the public's imagination because it's so easy to use. It was unveiled to the world just 11 days ago, and has already amassed more than a million users — gaining adoption more quickly than Facebook, which took ten months to hit the same milestone.

But there are challenges even the company behind it acknowledges, including the tendency to generate "nonsense" along the way.

Ask it anything

The prompts given to the bot can be silly, like asking it to write a movie script about elephants riding a roller coaster, or complex, like asking it to explain the history of the Middle East. It can write songs, law school essays, and even computer code.

Shouldn't this be in iambic pentameter? A poem written by the AI tool ChatGPT in response to the prompt: 'Write a poem about winter in the style of Shakespeare.' (Nisha Patel/CBC)

Momoh says the bot is better than anything that's come before at generating text responses to real human questions. He suggests people could use ChatGPT as a tool to enhance their productivity, especially in sectors like customer service, advertising and media.

"In a year or two, I think it will basically impact anything that involves generating text," he said.

While that may raise concerns about artificial intelligence putting people out of work, Melanie Mitchell, a computer scientist at the Santa Fe Institute, expects that jobs will just shift as workers are no longer required to complete repetitive tasks.

"Technology tends to create jobs in unexpected areas as it takes jobs away," she said.

'Incorrect or nonsensical answers'

The AI tool is in its early stages and users are discovering its limitations.

ChatGPT doesn't have a way to tell if the responses it's generating are true or false. Mitchell says that's a big problem, and for now, to use the bot for work responsibilities should require careful human fact-checking.

"I searched my own name and it said a lot of correct things about me, but it also said that I had passed away on Nov. 28, 2022, which is a little disturbing to read."

OpenAI has acknowledged the tool's tendency to respond with "plausible-sounding but incorrect or nonsensical answers," an issue it considers challenging to fix.

An ad jingle written by AI tool ChatGPT. (Nisha Patel/CBC)

Releasing ChatGPT to the public may help OpenAI find and fix flaws. While it's programmed the bot in an effort to avoid inappropriate tasks like asking for advice about illegal activities, or biased or offensive requests, some users have still found ways around the guardrails.

Because it's trained on existing language, AI technology can also perpetuate societal biases like those around race, gender and culture.

"So what's the effect that those answers could have? Maybe not on me or you, but again on a small child or somebody who's impressionable that is just trying to form their worldview on some of these harder topics," said Sheldon Fernandez, CEO of Darwin AI, which is working on harnessing AI for manufacturing.

Still some in the field like Fernandez are describing ChatGPT's debut as a "seminal" moment. And as the bot gains traction with the public, it's also sparking debate about when and how it should be used — and who should regulate it. 

"We need to think about that hard….One of the challenges with this is the technology just moves so quick and quicker than often legislative bodies can," said Fernandez.

Ask ChatGPT itself if the world is ready for it, and it spits out this answer:

"It is important for society to carefully consider these issues and develop a responsible approach to the use of AI technologies."


Nisha Patel is a senior business reporter with CBC News. She's been reporting on business and economics for more than a decade and has lived and worked in New York City, Edmonton and Calgary. She lives in Toronto with her husband and two kids. Find her on Twitter @nishapatel.

With files from Reuters

Add some “good” to your morning and evening.

A variety of newsletters you'll love, delivered straight to you.

Sign up now