Reflection on week 6: Artificial Intelligence, Academic Integrity, and more
This week we took a deeper look at AI…
Lucas Wright spoke on the widely used fast growing use of AI in the education system in Canada. He says how teachers, and students need to adopt Critical Generative AI Literacy. To do so, an AI user needs understand how the tools work and be aware of the implications of AI for society and the environment.
A big part of Lucas talk that stood out to me is how AI is advanced technology but has had flaws when it comes to accuracy
To explore this I test the accuracy of the following three prompts to AI:
Prompt: “Generate an image that looks like it was drawn by a kindergarten student of a green flamingo wearing a baseball cap”
Output:
Accuracy rating: 10/10
Prompt: “Act as a principal that has just found me stealing sugar packets from the cafeteria, you are a laid back chill principal and I am one of your favourite students”
Output:
[Leaning against the doorframe, arms crossed, half a smile]
Alright, alright, help me out here for a second… I walk into the cafeteria, minding my own business, and what do I see? One of my star students—someone I genuinely like—stuffing sugar packets into their hoodie like we’re running a black-market bakery
Accuracy rating: 10/10
Both the first prompts were very accurate so I wanted to try a more detail image to see if it would miss anything…
Prompt:
“Generate an image of three friends at a Japanese hotspring. The first friend is a chubby walrus wearing a green and blue polka dot bathing suit. The second friend is a hipster camel wearing a chain and white tank top. The third friend is a skinny deer wearing black glasses drinking a soda. Have nice polished wood in the background, as well as the hotsprings”
Output:
Accuracy: 9/10
Here we can see ChatGPT correctly creating my prompt but missing some accuracy where the walrus bathing suit does not really have green and I never said to give the camel glasses.