OpenAI’s chatbot, ChatGPT, launched in November and quickly went viral on social media. OpenAI’s CEO Sam Altman announced the bot on Twitter, saying that releasing the first edition of ChatGPT was a step closer to having language interfaces that act as “helpful assistants that talk to you, answer questions, and give advice.” People have already found that ChatGPT does just that, it works as a helpful assistant, answers a seemingly endless barrage of questions, and sometimes gives useless advice. ChatGPT isn’t just a language interface but also a neural network that learns as it goes. As people have played around with ChatGPT and its younger sibling ChatGPT-3, they have found some creative uses of the tech that will blow you away, from writing all the recipes for a customized thanksgiving dinner to detecting Alzheimer’s, writing books for children, and even, passing business school exams.
Before ChatGPT became accessible to the general public, New York Times writer, Priya Krishna, asked ChatGPT-3 to help her write the recipes for a Thanksgiving dinner based on some of the flavors she loves and memories from her previous Thanksgiving dinners. She wanted to see whether AI could take over the jobs of recipe developers. She started by providing the bot with information about her family background, preferences, and favorite ingredients. Then, she asked the bot to write a Thanksgiving menu tailored to her. Unfortunately, the recipes were not a hit, proving to Krishna that AI has a long way to go before it takes over her job as a recipe developer.
Researchers from Drexel University’s School of Biomedical Engineering, Science, and Health Systems found that the more advanced version of ChatGPT, ChatGPT3, can identify if a user is exhibiting early signs of Dementia 80% of the time. The bot picks up on clues hidden in the user’s spontaneous speech to identify whether the user displays the early signs of developing Dementia or Alzheimer’s. The researchers are investigating the possibility that natural language processing tools like ChatGPT3 can help detect or predict the development of Alzheimer’s.
60-80% of dementia patients experience language impairment, which is a significant early predictor of the development of Alzheimer’s. ChatGPT3 can pick up on subtle clues of language impairment, like forgetting the meaning of words, hesitation, and making grammar and pronunciation mistakes. Researchers hope that doctors could incorporate language processing exams into early screening for Alzheimer’s or dementia or even into routine visits. Early detection is crucial for treating Alzheimer’s, so the earlier indicators like language impairment are discovered, the better.
While ChatGPT can help researchers and scientists, it can also help the everyday person. For example, Ammarr Reshi used ChatGPT and Midjourney to write and illustrate an entire children’s book in 72 hours. Reshi came up with the idea to write a children’s book while reading one to his friend’s child. As a product-design manager at a Fintech company in San Francisco, he has little experience with creative writing or illustrating. So, he turned to ChatGPT to help him write the children’s book of his dreams. The book, “Alice and Sparkle,” is about a girl and her robot friend, Sparkle. He then used the AI image generator Midjourney to illustrate the book and self-published it on Amazon. While some people were impressed by Reshi’s creativity, others expressed anger, explaining that they felt Reshi had taken a shortcut to write and illustrate a book. Reshi told Business Insider that he wouldn’t call himself the author of the book, saying, “The AI is essentially the ghostwriter, and the other AI is the illustrator.” Reshi harnassed ChatGPT’s ability to generate creative long-form content and used it as a tool to achieve a lifelong goal.
ChatGPT’s ability to be creative helped it pass the Master of Business Administration final exam at Wharton University. The test was administered and graded by Professor Christian Terwiesch. Professor Terwiesch administered the test, asking ChatGPT to answer the questions and then authoring a paper about the experience called “Would ChatGPT Get a Wharton MBA? A Prediction Based on Its Performance in the Operations Management Course.” The bot scored between a B- and a B on the exam.
Terwiesch said that the bot did an amazing job with questions regarding basic operations management and process analysis questions. He also noted that ChatGPT made surprising mistakes on relatively simple mathematical calculations. Finally, ChatGPT struggled with the more advanced process analysis questions about process flows with multiple products or effects like demand variability. The implications for the passing grade are far-reaching. Terwiesch says in the paper that AI should be viewed as a tool to teach creative problem-solving and to improve teaching productivity. He also called for schools to focus on adapting exam policies and curricula to accommodate neural network-language-interface tools.
ChatGPT and ChatGPT3, can generate creative responses to complicated questions or long-form content. It can help scientists and doctors predict disease with its learning power. But according to professor Terwiesch, it fails to execute math problems at the sixth-grade level. So, we should probably wait for ChatGPT to get a bit better before we hand over the keys to the kingdom of the living. The many uses for ChatGPT as a language interface are just being discovered, and like Sam Altman said when he announced its launch, this is just the beginning of “what’s possible.”
.svg)
.svg)
