jasminebui1225
commented
1 year ago Decoding ChatGPT: A Conversation with Professor Daniel Linna on its Functionalities on WGN Radio
As we are 2 months into 2023, there is a shift away from traditional search engines towards the use of Large Language Model (LLM). And our outstanding mention of this conversation is ChatGPT developed by OpenAI. In a recent interview on WGN Radio, Professor Daniel Linna whose research is based in Northwestern University had shared his thoughts on how this technology works.
What is it?
The advanced model eliminates the cognitive load to decide whichever links to click on. By inputting a prompt such as questions or commands, ChatGPT gives you seamless, human-like responses within seconds. This versatile tool is your personalized Google search in many tasks.
Professor Linna repetitively mentioned the term “prediction” as ChatGPT is a trained model using a massive amount of text data from the Internet. The technology is to predict the next reasonable words to produce coherent sentences and paragraphs in a way that it can communicate with us effectively.
For instance, an example that was used in the dialogue:
- What comes after: Dog chases cat, cat chases (blank).
“Mouse!”
ChatGPT’s guesses are more complex and it is getting better as long as there is data streaming in.
How accurate are these responses?
It is tricky and essential to question its accuracy, especially when we associate the responses with facts. ChatGPT’s prediction has little to do with facts as it synthesizes the existing data to respond. This action of gathering information and mix-matching the words does not say the truth all the time, so it’s best to verify the answer and pay attention to details.
Is this everything LLM can offer to us?
Definitely not. ChatGPT is most well-known, but it only offers a partial portion of this technological realm. Professor Linna’s research on the use of computational LLM tools to expand access to legal matters & justice requires not only the answers to prompts, but also directions to solve legal problems. His envision is that this machine can take in our information and give us notions what steps to be taken. If written notes or letters are required, the innovation will handle your first draft. And this already exists when you think of Google Gmail and others, it analyzes your content and notifies whether you forgot to attach the files. Therefore, we still have a lot to look forward to.
Our concerns
ChatGPT provides a human-like interaction which crosses out the inflexibility that automatic responses would normally do. Nonetheless, without the right prompts, how can humans receive what they need? Amazingly, Professor Linna has been doing research with Professor Brunswicker who is also RCODI Director at Purdue University on these matters. They are studying chatbots with the intention to humanize these systems that can determine user problems faster and develop trust. And we are very excited for new updates on this collaborative research!
On the other hand, ChatGPT is confusing our education system when many students have access to this open tool and wrongly exploit it. It is our mission to quickly adapt and train students to utilize these new resources responsibly in creative ways.
For more details of this conversation, please check WGN Radio for a full version: Here
About Professor Daniel W Linna Jr.
Professor Daniel W Linna Jr. is a Senior Lecturer and Director of Law & Technology at Northwestern University, Illinois, United States. He is a specialist in intellectual property law and has written and taught on topics such as patent law, copyright law, trademark law, and trade secret law. You can find his biography here.
ChatGPT: An interview with John Williams about ChatGPT on WGNRadio: https://twitter.com/DanLinna/status/1619117132440080384.
Latest publications from Dr. Brunswicker & Satyam.
20 experts from around the world, group behavior and group including crowds using advanced. Aarhus workshop on experiments with Collectives/Groups.