From editing a work email to brainstorming ideas for a novel, chatbots are helping humans make conversations and ask different types of questions. OpenAI’s ChatGPT was the first chatbot which made people all around the world interact with AI models. After ChatGPT, several AI chatbots are helping users daily in different activities. The Washington Post looked at 200,000 English language conversations on two different chatbots which were built from the same technology as ChatGPT to find out what people really ask these chatbots in their day to day lives.
The Washington Post found that the most common questions users ask these AI chatbots are about creative writing. They ask AI chatbots to give them ideas about whatever fanfiction, lyrics, joke, poem or essay they are working on. Researchers say that AI chatbots are especially built for brainstorming so it is easier for people to brainstorm their ideas with the help of these AI models. The Post also found out that users also ask AI chatbots to suggest names for their business, character or book they are writing.
AI chatbots give thoughtful and fun responses when users push them for it. Many users try to ask AI chatbots to come up with random fantasy creatures and ideas and these chatbots actually deliver what the users are asking for. Almost one in six conversations are about students seeking help in their homeworks. Some students use these chatbots as tutors and try to understand the solutions of their problems from them.
Most AI chatbots are trained on publicly available data which includes textbooks, articles and historical writings. This makes it easier for them to give users a summarized version of their questions. But on the other hand, AI chatbots can also hallucinate and make up information while disguising it as a fact. Many educators are working on this rising AI problem and are checking their students’ homework or assignments by passing them through AI detectors.
5% of the conversations on these AI chatbots were about users asking personal questions like advice on flirting or how to tell if their partner is cheating. But AI chatbots are also prone to say some offensive or harmful things. Some users become personally attached to AI chatbots and listen to them well and follow whatever they say. They even give their full names, employer's names and other sorts of personal information to AI chatbots. Many experts have warned the users to not give their personal information to AI chatbots as the companies developing these chatbots often use the data to train their models.
7% of the conversations were also related to computers like writing, understanding or debugging computer codes. Computer students also use AI chatbots to write their coding assignments. This has raised warnings about future coding jobs as most of the work can be done with AI. But experts believe that AI chatbots aren't going to replace human coders.
15% of the conversations which were analyzed were about work related queries like drafting an email, writing presentations and doing e-commerce work while 2% of the conversations were asking AI chatbots to help find jobs or help users write résumé or a cover letter. Some were also asking AI chatbots to help them prepare for a job interview.
Even though the chatbots analyzed couldn't draw pictures, some users asked for image generation which the chatbots declined and asked users to take help from MidJourney(AI image generator). AI image generation bots can be helpful and creative but they can also create controversial images which could be biased or stereotypical.
13% of the AI prompts included the word “please” as most people have learned to interact with AI chatbots respectively. Some users used AI chatbots daily with one user having 13,213 conversations in just 201 days.
Read next:
• These Are The World’s Most Visited Websites In 2024
• AI Development Needs Double Microsoft's Electricity Use Between 2020 and 2023
The Washington Post found that the most common questions users ask these AI chatbots are about creative writing. They ask AI chatbots to give them ideas about whatever fanfiction, lyrics, joke, poem or essay they are working on. Researchers say that AI chatbots are especially built for brainstorming so it is easier for people to brainstorm their ideas with the help of these AI models. The Post also found out that users also ask AI chatbots to suggest names for their business, character or book they are writing.
AI chatbots give thoughtful and fun responses when users push them for it. Many users try to ask AI chatbots to come up with random fantasy creatures and ideas and these chatbots actually deliver what the users are asking for. Almost one in six conversations are about students seeking help in their homeworks. Some students use these chatbots as tutors and try to understand the solutions of their problems from them.
Most AI chatbots are trained on publicly available data which includes textbooks, articles and historical writings. This makes it easier for them to give users a summarized version of their questions. But on the other hand, AI chatbots can also hallucinate and make up information while disguising it as a fact. Many educators are working on this rising AI problem and are checking their students’ homework or assignments by passing them through AI detectors.
5% of the conversations on these AI chatbots were about users asking personal questions like advice on flirting or how to tell if their partner is cheating. But AI chatbots are also prone to say some offensive or harmful things. Some users become personally attached to AI chatbots and listen to them well and follow whatever they say. They even give their full names, employer's names and other sorts of personal information to AI chatbots. Many experts have warned the users to not give their personal information to AI chatbots as the companies developing these chatbots often use the data to train their models.
7% of the conversations were also related to computers like writing, understanding or debugging computer codes. Computer students also use AI chatbots to write their coding assignments. This has raised warnings about future coding jobs as most of the work can be done with AI. But experts believe that AI chatbots aren't going to replace human coders.
15% of the conversations which were analyzed were about work related queries like drafting an email, writing presentations and doing e-commerce work while 2% of the conversations were asking AI chatbots to help find jobs or help users write résumé or a cover letter. Some were also asking AI chatbots to help them prepare for a job interview.
Even though the chatbots analyzed couldn't draw pictures, some users asked for image generation which the chatbots declined and asked users to take help from MidJourney(AI image generator). AI image generation bots can be helpful and creative but they can also create controversial images which could be biased or stereotypical.
13% of the AI prompts included the word “please” as most people have learned to interact with AI chatbots respectively. Some users used AI chatbots daily with one user having 13,213 conversations in just 201 days.
Read next:
• These Are The World’s Most Visited Websites In 2024
• AI Development Needs Double Microsoft's Electricity Use Between 2020 and 2023
