AI Made Four of These. Can You Guess Which Ones?

Code a website. Translate French. Explain how democracy works to a toddler. Write song lyrics in the style of Ed Sheeran.

The new kid in town can do all of this—and more—in seconds. But it’s not a kid at all. It’s a chatbot called ChatGPT, and some people say it represents the biggest advancement in technology in decades. 

Chatbots are computer programs made to imitate human conversation. Many, including ChatGPT, are powered by artificial intelligence (AI). That’s a type of technology that lets machines do things that normally need a human’s ability to think or learn, such as understand language. 

If you’ve ever asked Siri for the weather or told Alexa to play a song, then you’ve spoken with basic AI devices, known as virtual assistants. But chatbots are taking AI to a whole new level—with ChatGPT leading the way. It’s the most advanced chatbot ever made available to the public, tech experts say. 

ChatGPT can carry on lengthy, humanlike discussions. It can answer questions—and explain its reasoning. It can also respond to feedback. If you don’t understand its explanation of algebra, for example, ChatGPT will make it simpler until you do.

Code a website. Translate French. Teach a toddler how democracy works. Write song lyrics in Ed Sheeran’s style.

The new kid in town can do all of this and more. In seconds! But it is not a kid at all. It is a chatbot called ChatGPT. Some people say it represents the biggest advancement in technology in decades.

Chatbots are computer programs that have been designed to imitate human conversation. Many, including ChatGPT, are powered by artificial intelligence (AI). AI is a type of technology that lets machines do things that normally need a human’s ability to think or learn. One example is understanding language.

Have you ever asked Siri for the weather or told Alexa to play a song? If so, you have spoken with basic AI devices. They are called virtual assistants. But chatbots are taking AI to a whole new level. ChatGPT is leading the way. Tech experts say it is the most advanced chatbot ever made available to the public.

ChatGPT can carry on long, humanlike discussions. It can answer questions—and explain its reasoning. It can also respond to feedback. For example, say you do not understand how it explained algebra. ChatGPT will make it simpler until you do.

More than 100 million people have used ChatGPT since a California company called OpenAI released a free version to the public last year. Other tech leaders are racing to develop equally powerful AI tools. These new breakthroughs can do more than just chat. AI systems like Dall-E and Canva create images and artwork based on text prompts. 

These advances have experts questioning how this new technology could change our lives. Will AI replace human workers? Help diagnose diseases? Spread false information? 

One thing is certain: The new AI tools represent a major turning point. They could transform people’s lives as much as the telephone and the internet did (see “Tech Over Time,” below).

In an online post, Aaron Levie, chief executive of the tech company Box, summed it up this way: “ChatGPT is one of those rare moments in technology where you see a glimmer of how everything is going to be different going forward.”

OpenAI is a California company. Last year, it released a free version of ChatGPT to the public. Since then, more than 100 million people have used it. Other tech leaders are racing to develop equally powerful AI tools. These new tools can do more than just chat. AI systems like Dall-E and Canva can create images and artwork based on text prompts.

Experts are questioning how this new technology could change our lives. Will AI replace human workers? Help identify diseases? Spread false information?

One thing is for sure: The new AI tools represent a major turning point. They could change people’s lives as much as the telephone and the internet did (see “Tech Over Time,” below).

Aaron Levie is chief executive of Box, a tech company. In an online post, he said, “ChatGPT is one of those rare moments in technology where you see a glimmer of how everything is going to be different going forward.”

So how do ChatGPT and other chatbots, such as Google’s Bard, work? They are powered by a kind of artificial intelligence called a neural network. 

A neural network is a mathematical system that learns skills by analyzing huge amounts of digital data. After examining thousands of tiger photos, for example, a neural network learns to recognize a tiger.

Around 2018, companies like Google and OpenAI began building neural networks that learned from massive amounts of internet text, including books, Wikipedia entries, chat logs, and more. 

These neural networks—including ChatGPT—are known as large language models. They use tons of internet data to create a map of human language. Then they use that map to perform different tasks, from generating computer programs to holding a conversation.

If you ask ChatGPT about George Washington’s birthdate, for example, it refers to its language map to find how those words fit together. Then it responds: “George Washington was born on February 22, 1732.”

So how do ChatGPT and other chatbots, such as Google’s Bard, work? They are powered by a kind of artificial intelligence called a neural network.

A neural network is a mathematical system. It learns skills by analyzing huge amounts of digital data. For example, after examining thousands of photos of tigers, it learns to recognize a tiger.

Around 2018, companies like Google and OpenAI began building neural networks. Those networks learned from huge amounts of internet text, like books, Wikipedia entries, chat logs, and more.

ChatGPT and other neural networks are known as large language models. They use tons of internet data to create a map of human language. Then they use that map to do different tasks. Such tasks could include creating computer programs or holding a conversation.

For example, if you ask ChatGPT about George Washington’s birth date, it checks its language map to find how those words fit together. Then it responds: “George Washington was born on February 22, 1732.”

Though AI technology is still relatively new, many people have high hopes that it could make the world better in big and small ways. In our everyday lives, AI could take over tasks we don’t like doing, such as searching the internet for the best prices. Companies are building bots that can use websites the same way a human does. In their next stage, AI systems could shop online for your favorite jeans or track how you spend your allowance. 

What about helping to solve some of our most serious problems? Because of its ability to analyze data and images quickly, AI could help us detect diseases and discover new medical treatments. AI was recently used to help a paralyzed stroke survivor talk again by translating her brain signals into words spoken by a digital avatar. 

AI technology is still quite new. But many people have high hopes that it could make the world better in big and small ways. In our daily lives, AI could take over tasks we do not like, such as searching the internet for the best prices. Companies are building bots that can use websites the same way a human does. In their next stage, AI systems could shop online for your favorite jeans. Or they could track how you spend your allowance.

What about helping to solve some of our most serious problems? AI can analyze data and images quickly. That means it could help us detect diseases and discover new medical treatments. Recently, AI was used to help a paralyzed stroke survivor talk again. It did so by translating her brain signals into words spoken by a digital avatar.

An advanced version of ChatGPT is now assisting people with vision loss. The AI analyzes photos and explains aloud what’s in them. It can describe photos of streets and rooms—and get even more creative. If a user uploads an image of the inside of their refrigerator, the AI can suggest recipes for dinner using the food in the fridge.

In education, AI tools may soon act as personal tutors. Khan Academy, an education nonprofit, is testing an AI bot that helps students practice debating, writing, and taking quizzes. The bot can even simulate historical figures like Benjamin Franklin and Harriet Tubman, allowing kids to “chat” with them.  

An advanced version of ChatGPT is now helping people with vision loss. The AI analyzes photos. Then it explains aloud what is in them. It can describe photos of streets and rooms. And it can get even more creative. If a user uploads an image of the inside of their refrigerator, the AI can suggest dinner recipes using the food in the fridge.

In education, AI tools may soon act as personal tutors. Khan Academy is an education nonprofit. It is testing an AI bot that helps students practice debating, writing, and taking quizzes. The bot can even simulate historical figures like Benjamin Franklin and Harriet Tubman. It allows kids to “chat” with them.

If AI can do so much, what will be left for people to do? That’s one of the concerns about the technology. 

For now, the new AI tools are too unreliable to replace human workers. The bots were trained on internet text—which isn’t always factual—so they sometimes get things wrong and even make stuff up. Their responses can be offensive and full of harmful stereotypes absorbed from the internet. (Researchers are trying to fix these issues.) AI also can have trouble generating images of certain things—such as human hands. Plus, ChatGPT’s training ended in 2021—it has no information on anything that has happened since.

Still, AI can help people do some jobs faster. One study found that computer coders using an AI tool completed their tasks 56 percent faster than those who did not.

If AI can do so much, what will be left for people to do? That is one of the concerns about the technology.

For now, the new AI tools are too unreliable to replace human workers. The bots were trained on internet text. That is not always factual. So sometimes they get things wrong and even make stuff up. Their responses can be offensive and full of harmful stereotypes taken from the internet. (Researchers are trying to fix these issues.) AI also can have trouble creating images of certain things, like human hands. Plus, ChatGPT’s training ended in 2021. That means it has no information on anything that has happened since.

Still, AI can help people do some jobs faster. One study found that computer coders using an AI tool finished their tasks 56 percent faster than those who did not.

As the AI systems get more sophisticated, they will likely replace human workers in some professions. The McKinsey Global Institute estimates that 12 million people may need to change jobs by 2030 because of AI advances. Workers who do repetitive tasks—like customer service agents and telemarketers—are most likely to be affected first. 

Typically, when jobs are taken over by machines, new jobs are also created. But it’s too early to know exactly how many of even those new jobs AI could impact. 

“Historically, if your job gets automated, you find something new,” says Harry Holzer, an economist at Georgetown University in Washington, D.C. “With AI, the thing that’s kind of scary is it could simply grow and take over more tasks.” 

Some people are determined not to let that happen. Last May, for example, TV and movie writers went on strike partly to demand rules about AI. They don’t want studios to use bots instead of humans to write scripts. Actors, who later also went on strike, worry that AI-generated replicas could someday replace them onscreen. 

As the AI systems get more sophisticated, they will likely replace human workers in some types of work. The McKinsey Global Institute estimates that 12 million people may need to change jobs by 2030 because of advances in AI. Workers who do repetitive tasks are most likely to be affected first—customer service agents and telemarketers, for example.

Usually, when jobs are taken over by machines, new jobs are also created. But it is too early to know exactly how many of even those new jobs AI could impact.

“Historically, if your job gets automated, you find something new,” says Harry Holzer. He is an economist at Georgetown University in Washington, D.C. “With AI, the thing that’s kind of scary is it could simply grow and take over more tasks.”

Some people are determined not to let that happen. Last May, for example, TV and movie writers went on strike. One reason was to demand rules about AI. They do not want studios to use bots instead of humans to write scripts. Actors later also went on strike. They worry that AI-generated copies could someday replace them onscreen.