Why do you think the problems of AI bias and representation in data sets was not seen immediately in the existing datasets collected by researchers and companies? Give some specific reasons why this was overlooked and some examples. (1 paragraph, 5-7 sentences)
March 25
12m
C
What was the mechanical Turk in your own words. What is an example of a “mechanical Turk” in today’s society? (1 paragraph, 5-7 sentences)
https://en.wikipedia.org/wiki/Mechanical_Turk
Based off this version of mechanical turk.
BlackBoxes Video Including alexa and mechanical turk.
March 25
10m
C
Design a cybernetic system diagram for a future AI system (not one that exists now) that would support human mental health in some way. We want to see you extrapolate from the history of cybernetics and current AI technologies to imagine a possible future implementation of a cybernetic system. Don’t be afraid to take some creative leaps here. However, this system should be justified both from a technical perspective, but also drawing on the extended literature of cybernetics.
Not allowed: magic, infinite energy sources, terminator robots (or other familiar tropes from science fiction)
Allowed: brain machine interfaces, human in the loop, faster processing, new sensors that extract more information from the physical world than is currently possible.
We want you to teach us something new about cybernetics!
Step 1: Draw the basic building blocks of your system with arrows showing the signal flow. Make sure to include enough technical description to describe the inputs/outputs, their dimensions/modality (i.e. text data, images, video, haptic forces, etc)
Step 2: Annotate each block with a short description identifying which cybernetics concept it illustrates (e.g. feedback, homeostasis, entropy, information, etc). Make sure to include quotes and references from Wiener or other cybernetics literature.
Step 3: Describe the emergent properties of your system. What makes it more than the sum of its parts? What are potential dangers or ethical considerations of your system deployed for individual or community mental health?
March 25
Go to https://teachablemachine.withgoogle.com/train Links to an external site. and train your own model. Show a screenshot of your trained model and write a paragraph about what task did you train your model to do. Did you find the learned model was successful?
https://teachablemachine.withgoogle.com/train
Link for assignment.
March 25
9m
C
Step 1: Read the paper on dataset nutritional labels
Links to an external site.
Step 2: Pick an example dataset of your choosing. You can use tools like Kaggle.com, or Google’s Dataset Search
Links to an external site. to find datasets. It might be useful to find datasets related to machine learning or deep learning, these are typically larger in nature.
Step 3: Fill out the Datasheet questions using the template provided here: LINK
.
Step 4. Create a dataset nutritional label for your dataset. You can look at these examples to guide your example: NYC Property Value Dataset
Links to an external site., NYC Tax Bill Dataset Links to an external site., Melanoma Dataset
Links to an external site.
To help create your dataset, you can use the following template: Template
To be submitted: (1) your datasheet for your dataset (2) your nutritional label filled out.
Those three go with that 1
Neal Stephenson admitted that Snow Crash is actually a satire. Pick one aspect of the story from this week’s reading and explain what/who it is satirizing, and whether the satire is effective. Posts should be 2 paragraphs long and reference particular chapters/page numbers in the reading. [7 points] March 25.
All of the Snow Crash ones are from a book
thats all of them for the 25
Last Completed Projects
| topic title | academic level | Writer | delivered |
|---|