The Chatbot Is Speaking

What does ChatGPT—the buzzy artificial intelligence (A.I.) chatbot—have to say when asked to generate a dramatic one-sentence opening for a news article about how ChatGPT might change the world?

As the world stands on the brink of a new age of artificial intelligence, ChatGPT rises as a revolutionary force, poised to change the way we think, communicate, and live our lives.

And on whether we humans should be worried about this new leap in A.I. capability?

As ChatGPT becomes more advanced, the question on everyone’s mind is whether this AI will bring about a utopia or a dystopia for humanity.

ChatGPT is, quite simply, the best artificial intelligence chatbot ever released to the general public, according to New York Times technology columnist Kevin Roose. Built by the San Francisco-based company OpenAI, ChatGPT—the “GPT” stands for generative pre-trained transformer—gained more than a million users in less than a week after its November 30 release. It’s made headlines everywhere for just how good it is at responding to questions and commands. Scary good, according to some.

ChatGPT is an artificial intelligence (A.I.) chatbot. What does the chatbot say when asked to write a one-sentence opening for a news article about how ChatGPT might change the world?

As the world stands on the brink of a new age of artificial intelligence, ChatGPT rises as a revolutionary force, poised to change the way we think, communicate, and live our lives.

And about whether we humans should be worried about A.I.?

As ChatGPT becomes more advanced, the question on everyone’s mind is whether this AI will bring about a utopia or a dystopia for humanity.

ChatGPT is the best artificial intelligence chatbot ever released to the general public. That’s according to New York Times technology columnist Kevin Roose. It was created by the San Francisco-based company OpenAI. The “GPT” stands for generative pre-trained transformer. ChatGPT gained more than a million users in less than a week after its November 30 release. It’s made headlines everywhere because it is good at responding to questions and commands. Scary good, according to some.

ChatGPT can write jokes, debug computer code, compose music, draft contracts, and create college-level essays. It can explain a complex scientific topic to an expert—or to a 5-year-old. It can generate responses in the style of a pirate or a Taylor Swift song. Want an article summary, a bullet list, or an answer to a tricky question? It can generate all this in a matter of seconds. Many are speculating it could replace Google as a conventional search engine.

Experts say it will change the world for both the better and the worse. It could make life easier by improving access to information and automating certain tasks, but it’s so good at some tasks that it might replace human workers and eliminate jobs. Many are also concerned it will spread misinformation and perpetuate harmful biases.

“ChatGPT is one of those rare moments in technology where you see a glimmer of how everything is going to be different going forward,” wrote Aaron Levie, CEO of the cloud storage company Box, in a December tweet. Levie is one of the many executives exploring the ways ChatGPT and similar chatbots will change the business landscape.

ChatGPT can write jokes. It can debug computer code. It can compose music. It can draft contracts. And it can create college-level essays. It can also explain a complex scientific topic to an expert or to someone who is 5 years old. It can respond as a pirate or a Taylor Swift song. Do you want an article summary, a bullet list, or an answer to a tricky question? It can generate all this in a matter of seconds. Many say it could replace Google as a standard search engine.

Experts say ChatGPT will change the world for both the better and the worse. It could make life easier by improving access to information and automating certain tasks. ChatGPT is very good at some tasks. It might replace human workers and eliminate jobs. Many are also concerned it will spread misinformation and harmful biases.

“ChatGPT is one of those rare moments in technology where you see a glimmer of how everything is going to be different going forward,” wrote Aaron Levie in a tweet in December. Levie is the CEO of the cloud storage company Box. Levie is one of the many executives studying the ways ChatGPT and similar chatbots will change the business landscape.

ChatGPT is what scientists call a “large language model.” It works by analyzing enormous amounts of digital text, including books, Wikipedia articles, news stories, and online chat logs. It identifies billions of distinct patterns in the way people connect and relate words, numbers, and symbols.

Such A.I. text systems aren’t new. Google uses large language models in its search engine. A.I. chatbots aren’t new either. You spoke to a bot the last time you asked Siri about the weather. And you might have spoken to a bot when you last chatted with customer service about returning an item online.

ChatGPT and similar technologies such as Google’s Language Model for Dialogue Applications (LaMDA) are what researchers call “neural networks,” or mathematical systems loosely modeled on the network of neurons in the human brain. This is the same technology that translates between French and English on Google Translate and identifies pedestrians as self-driving cars navigate city streets.

“You now have a computer that can answer any question in a way that makes sense to a human,” Levie says. “It can extrapolate and take ideas from different contexts and merge them together.”

Scientists call ChatGPT a “large language model.” It works by analyzing enormous amounts of digital text. That includes books, Wikipedia articles, news stories, and online chat logs. It identifies billions of different patterns in the way people connect and relate words, numbers, and symbols.

These A.I. text systems aren’t new. Google uses large language models in its search engine. A.I. chatbots aren’t new either. Last time you asked Siri about the weather, you spoke to a bot. And you might have spoken to a bot when you chatted with customer service about returning an item online.

Researchers call ChatGPT and similar technologies, such as Google’s Language Model for Dialogue Applications (LaMDA), “neural networks.” They are mathematical systems loosely based on the network of neurons in the human brain. This is the same technology that translates between French and English on Google Translate. It’s the same technology that recognizes pedestrians when self-driving cars drive in city streets.

“You now have a computer that can answer any question in a way that makes sense to a human,” Levie says. “It can extrapolate and take ideas from different contexts and merge them together.”

Generative A.I. first captured the public’s attention last spring when OpenAI released a system called DALL-E. It lets people generate new images simply by describing what they want to see. DALL-E’s release kick-started what many are calling an A.I. gold rush. In 2022, investors pumped at least $1.37 billion into generative A.I. companies, according to data from PitchBook, which tracks financial activity across the industry.

Many hope ChatGPT will make life easier and think it could be a tool for good. Companies are already working to incorporate it into their products. One day, sooner rather than later, it might be used to get on-demand access to information from medical professionals or as a personal psychotherapy bot.

Ben Whittle, a landscaper in rural England with dyslexia, uses ChatGPT to help craft error-free emails to new clients. Several people on TikTok with ADHD have posted about using ChatGPT as a personal tutor or to help generate lists and schedules.

But many fear that people are relying too much on the bot too soon. Even leaders in A.I. admit the technology still has a long way to go. OpenAI’s CEO, Sam Altman, acknowledged the shortcomings of ChatGPT in a December 10 tweet.

“ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness,” he wrote. “It’s a mistake to be relying on it for anything important right now.”

Generative A.I. captured the public’s attention last spring. That’s when OpenAI announced a system called DALL-E. It lets people generate new images by describing what they want to see. DALL-E’s release began what many are calling an A.I. gold rush. In 2022, investors put at least $1.37 billion into generative A.I. companies. That’s according to data from PitchBook, which tracks financial activity across the industry.

Many hope ChatGPT will make life easier. ChatGPT could be a tool for good. Companies are already working to incorporate it into their products. One day soon, it might be used to get on-demand access to information from medical professionals or as a personal psychotherapy bot.

Ben Whittle is a landscaper in rural England with dyslexia. He uses ChatGPT to help write emails with no errors to new clients. Several people on TikTok who have ADHD have posted about using ChatGPT as a personal tutor. Some use ChatGPT to help generate lists and schedules.

But many think that people are relying too much on the bot too soon. Even leaders in A.I. say the technology still has a long way to go. Sam Altman is the CEO of OpenAI. He acknowledged the shortcomings of ChatGPT in a December 10 tweet.

“ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness,” he wrote. “It’s a mistake to be relying on it for anything important right now.”

The technology can sometimes generate responses that sound correct but are, in reality, nonsensical or incorrect—or even just slightly off. Certain problems or questions relating to math and physics can be particularly tricky for the bot. Plus, it’s only been trained on data before the year 2021, so it draws a blank when asked about anything current.

Many also say it will never be able to replace human creativity (see “Can a Chatbot Write a Sonnet?” below). After all, ChatGPT doesn’t exactly create anything new. Instead, it generates new content by feeding off gigantic portions of the internet and regurgitating patterns.

Some are also concerned about bots’ potential to spread misinformation and perpetuate harmful biases. In fact, this is an ongoing issue for A.I. Seven years ago, Microsoft released a chatbot called Tay, but it soon began spewing racist and xenophobic language, forcing the company to remove it from the Internet. Last year Facebook’s parent company, Meta, took down a newer chatbot for similar reasons.

The technology can sometimes generate responses that sound correct. But in reality they don’t make sense, are incorrect, or even just slightly off. Certain problems or questions relating to math and physics can be tricky for the bot. ChatGPT has been trained on data before the year 2021. It doesn’t know about anything current.

Many say ChatGPT will never be able to replace human creativity (see “Can a Chatbot Write a Sonnet?” below). ChatGPT doesn’t exactly create anything new. ChatGPT generates new content by taking large portions of the internet and repeating its patterns.

Some are also concerned that bots can spread misinformation and biases. This is an ongoing issue for A.I. Seven years ago, Microsoft released a chatbot called Tay. Tay soon began saying racist and xenophobic words. That forced the company to remove Tay from the Internet. Last year, Meta, Facebook’s parent company, took down a new chatbot for the same reasons.   

Following ChatGPT’s release, a flurry of concern about what it could mean in education swept through schools. Some experts and educators have predicted that ChatGPT, and tools like it, could spell the end of homework and take-home assignments.

“We’re at the beginning of a societal transformation,” says Brian Christian, a computer scientist and author of The Alignment Problem, a book about ethical concerns surrounding A.I. systems. “There’s going to be a bigger question here for businesses, but in the immediate term, for the education system, what’s the future of homework?”

New York City, Los Angeles, and Baltimore public schools are among those that have banned access to ChatGPT on school computer networks. And some educators have already changed the way they do things in the classroom following ChatGPT’s public release.

After reading a paper composed by a student using ChatGPT, Northern Michigan University Professor Antony Aumann decided to change essay writing for this semester: Students are now required to write first drafts of their essays in the classroom, using browsing monitors that restrict computer activity, and in later drafts, students have to explain each revision.

Tools to combat A.I.-generated plagiarism—and crack down on ChatGPT use in school—are also in the works. Princeton University senior Edward Tian spent part of his winter break creating GPTZero, an app that detects whether text was written by a human or A.I. (see “Policing ChatGPT” below).

There’s no doubt that ChatGPT marks a turning point in technology. Many have compared its release to the release of the first iPhone in terms of how it will transform our lives. Experts everywhere—including ChatGPT itself—say it will reshape the world in years to come.

When asked to provide a dramatic one-sentence final quote for a news article about the future of ChatGPT, here’s what the A.I. chatbot had to say:

As ChatGPT evolves, it holds the power to shape our very reality, but it’s up to us to wield it with responsibility and caution, for the fate of humanity may depend on it.

After ChatGPT was released, schools were concerned about its effect on education. Some experts and educators have said that ChatGPT, and tools like it, could bring homework and assignments at home to an end.

“We’re at the beginning of a societal transformation,” says Brian Christian. He’s a computer scientist and author of The Alignment Problem, a book about ethical concerns about A.I. systems. “There’s going to be a bigger question here for businesses, but in the immediate term, for the education system, what’s the future of homework?”

Public schools in New York City, Los Angeles, and Baltimore are among those that have banned access to ChatGPT on school computer networks. And some educators have already changed the way they do things in the classroom since ChatGPT was released.

Antony Aumann is a professor at Northern Michigan University. He read a paper composed by a student who used ChatGPT. After he read it, Aumann changed essay writing for this semester. His students need to write first drafts of their essays in the classroom. They must use monitors that restrict computer activity. In later drafts, students must explain each revision.

Tools to combat A.I.-generated plagiarism—and stop the use of ChatGPT in school—are been planned. Edward Tian, a Princeton University senior, spent part of his winter break creating GPTZero. GPTZero is an app that can say if a text was written by a human or A.I. (see “Policing ChatGPT” below).

There’s no doubt that ChatGPT is a change in technology. Many have compared its release to the release of the first iPhone. That’s because of how it could change our lives. Experts everywhere—and ChatGPT too—say it will change the world in years to come.

ChatGPT was asked to provide a dramatic one-sentence final quote for a news article about the future of ChatGPT. Here’s what the A.I. chatbot had to say:

As ChatGPT evolves, it holds the power to shape our very reality, but it’s up to us to wield it with responsibility and caution, for the fate of humanity may depend on it.

With reporting from Cade Metz, Kevin Roose, and Kalley Huang of The Times.

With reporting from Cade Metz, Kevin Roose, and Kalley Huang of The Times.