• Alone or with a parter, come up with a p5.js sketch that prompts the user for some sort of input using their natural language - and then uses the input in a meaningful and creative way. Use the Prompting sketch as a starting point. (An example could be asking users how they felt, and using ChatGPT to “understand” and classify the free text input. And then use this to generate related visuals, or similar.)

<aside> <img src="/icons/chat_lightgray.svg" alt="/icons/chat_lightgray.svg" width="40px" /> Click here to see the sketch

For the time issue: The biggest reason could be that all the value (the color, size, position, numbers of shapes) were generated by OpenAI, because I want each of them to be totally deferent instead of simply let AI fill colors or change one thing for me. So it takes a much longer time to respond to all my requests.

After I learned from others’ brilliant sketches, I find that I should give a rough framework (code) to minimize the AI's workload.

</aside>

• What could be a second, more ambitious version of the same project (or a different idea) that you wouldn’t know how to tackle (yet)?

<aside> <img src="/icons/chat_lightgray.svg" alt="/icons/chat_lightgray.svg" width="40px" /> ▶︎ make the expression more vivid with moving objects or interactions between the user and the facial expression ▶︎ add more elements more than circle (I tried arc for the mouse but it didn’t show at all (T^T) ▶︎ shorten the waiting time, or label it out when the openAI is still processing, to let the user be clear about whether the requirement is valid or not

</aside>

• Come up with 3-5 questions about anything relating to week 5’s content? (WebAPIs, how to use ChatGPT etc)

<aside> <img src="/icons/chat_lightgray.svg" alt="/icons/chat_lightgray.svg" width="40px" /> Q1: What are the factors that determine the 3D coordinates of a word in embedding？How the embeddings help? Q2: When asking chatGPT questions, someone told me to be as specific as possible. but sometimes, I find the answer not as satisfying as the one I got from a broad question. How can I strike the balance between them? Q3: ChatGPT will crash out of too many people are using it at the same time. How can this be solved? While many people using ChatGPT 3.5, will it affect the use of ChatGPT4?

</aside>

<aside> <img src="/icons/chat_lightgray.svg" alt="/icons/chat_lightgray.svg" width="40px" /> Q1: I was once told that ChatGPT has a memory, but there is a limit to how long it can remember, so the user needs to keep repeating the same requests. When playing a similar game, how can I ensure that it keeps roleplaying according to the original instructions and rules？ Q2: Sometimes, when I ask ChatGPT questions, it generates answers pretty slowly. If similar technology is applied to NPCs in the game world, how could the game designer make sure the communication between users and NPCs is smooth enough to ensure a better experience? (OMG above is about week6. I go through the wrong slides)

</aside>