E57 – ChatGPT

I. Introduction to ChatGPT

Brief overview of what ChatGPT is and how it works
ChatGPT (Generative Pre-trained Transformer)[1] is a chatbot launched by OpenAI in November 2022. It is built on top of OpenAI’s GPT-3.5 family of large language models, and is fine-tuned with both supervised and reinforcement learning techniques.
Reached 1 million subscribers users in a week.
An OpenAI blog post states that their ChatGPT model is trained with a deep machine learning technique called Reinforcement Learning from Human Feedback (RLHF) that helps simulate dialogue, answer follow-up questions, admit mistakes, challenge incorrect premises and reject inappropriate requests.
Setting the stage for the discussion of the pros and cons
- Clearly this will have impact in our communications in the future. Will the benefits outweigh the disadvantages? Let’s take a look

II. Pros of ChatGPT

III. Cons of ChatGPT

Limited to providing pre-determined responses
Difficulty handling more complex or nuanced questions
Lack of personalization and human touch
It makes one question the content being provided. Not original or human made?
The school system will need to respond to schoolwork, papers. Will it change us forever?
Potential security issues? Spam or virus generation

IV. Balancing the pros and cons: when is ChatGPT appropriate to use?

Situations where ChatGPT can be a valuable tool (e.g. customer service, FAQs)
Situations where human interaction is still necessary (e.g. sales, personalized support)

V. Conclusion and future outlook for ChatGPT

Recap of the main points discussed
Thoughts on the potential future developments and evolution of ChatGPT technology

The UC Moment : ChatGPT is here and prepare yourself.