A recently released AI-powered chatbot called ChatGPT launched this week to a mixture of praise and concern. Developed by OpenAI, the chatbot can teach users various things, like setting up a website, but it has also allegedly proven problematic at the same time.
The chatbot was built from the ground up to be as natural as possible when talking to people in what is called "a conversational style."
The chatbot's website says that "the dialogue format allows ChatGPT to answer follow-up questions, admit mistakes, question wrong premises, and turn down inappropriate requests."
Even though ChatGPT is not a new technology, many technologists have already been pleased with its ability to mimic human speech patterns and language while providing precise and timely information.
Many people have taken to social media to share their odd, sometimes amusing, and sometimes both interactions with the bot.
Cleo Abram, a video journalist, tweeted a video of her requesting the bot to “explain nuclear fusion in the style of a limerick," along with the statement, "I’m finding my biggest limitation to use it is *my* imagination!”
testing ChatGPT. this is one of the most interesting and impressive tools I’ve seen in a long time (the model and how you can use it now)
I’m finding my biggest limitation to use it is *my* imagination! what should I try asking it to do? pic.twitter.com/ZmSmfmeT5k
— Cleo Abram (@cleoabram) December 2, 2022
Another new user, Jeff Yang, asked the chatbot to “explain zero point energy but in the style of a cat.” The answer was funny and, to be honest, very creative for a complex algorithm.
ChatGPT really sold it with that final Meow! pic.twitter.com/np3UKpemFf
— Jeff Yang (@originalspin) December 2, 2022
Some people thought that Google might no longer be the best search engine because of how well the chatbot did early on.
TechCrunch managing editor Darrell Etherington said the ChatGPT search inquiries were as basic as if a user "were slacking with a colleague or interacting with a customer support agent on a website."
Etherington showed the power of the chatbot by asking a question about Pokemon and their strengths and weaknesses. “The result is exactly what I’m looking for—not a list of things that can help me find what I’m looking for if I’m willing to put in the time, which is what Google returns,” he explained.
All very interesting, but other technologists are concerned about the chatbot's potential for nefarious use. For example, in theory, it could teach someone how to build a bomb, etc.
Samczsun, a research partner and head of security at Paradigm, a financial institution that backs crypto and Web3 businesses, tweeted that he had managed to get past the chatbot's content filter. In a tweet, he showed evidence purportedly demonstrating he had managed to persuade the bot to teach him how to make a Molotov cocktail.
According to NBC News, this image has been verified as genuine by other spokespeople at Paradigm.
When testing an AI's safety and content filters, researchers and programmers frequently ask questions like how to manufacture Molotov bombs and hot-wire cars. Some people said they had tricked the bot into telling them how to make a nuclear bomb.
Even though there are safeguards to stop ChatGPT from responding to dangerous requests, OpenAI says on its website that the technology is not perfect.
"While we’ve made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior," a statement on OpenAI's website reads. It says that OpenAI is using moderation tools to prevent inappropriate responses, but "we expect it to have some false negatives and positives for now."
The website cautions that ChatGPT occasionally provides absurd or mistaken responses, even if the answers appear correct.
OpenGPT is not the first popular artificial intelligence from OpenAI to gain popularity. DALL-E, a machine that could create images from basic written instructions, became popular in 2021. DALL-E highlighted advancements in artificial intelligence's ability to learn in humanlike ways. However, that version of the AI and another one called DALL-E 2 came under fire for their racial and gender bias.
there is a lot more demand for ChatGPT than we expected; we are working to add more capacity
— Sam Altman (@sama) December 1, 2022
The demand for ChatGPT was so great on Thursday that OpenAI CEO Sam Altman tweeted that the business was doing everything possible to accommodate users.
“There is a lot more demand for ChatGPT than we expected; we are working to add more capacity,” Altman wrote.
In a follow-up tweet, Altman added: “also, it makes all of us at openai so happy to see people enjoying chatgpt so much, and doing such creative things!”
SHOW COMMENT (1)
Norman Wagner from the University of Delaware tells Interesting Engineering about the challenges of making extraterrestrial cement for off-space infrastructure.
Photos: The technology turning agricultural waste into 'sticky' bio-oil for permanent carbon storage
Quantum in 2027: Take a quantum leap into the future of ITIn a first, researchers discovered a rare mineral that comes directly from Earth's lower mantle
Sorry hydrogen, electric cars have already won — here's why
