AI in College - Faculty
What goes in is what comes out!
GAITs were trained mainly on information from the open internet. Many of those sources are biased, and GAITs may magnify that bias.
Additionally, any bias in the human-provided prompt results in a biased response from the GAIT.
- Reflections before the storm: the AI reproduction of biased imagery in global health visuals"We propose that generative artificial intelligence (AI), which uses real images as a basis for learning, might further serve to show how deeply embedded the existing tropes and prejudices are within global health images. This in turn can perpetuate oversimplified social categorisations and power imbalances."
What's your privacy worth?
Tech companies collect user input to further train the GAITs. Many tech companies do not allow their employees to use GAITs for work, out of fear that confidential information will be compromised.
Recently, Zoom had to revise their terms and conditions involving data collection for training their GAITs. It appeared that Zoom would collect all data from meetings on their platform, and use it to train their GAIT. There was such an outcry over the implied breach of privacy for schools and corporations that Zoom walked back that policy.
It is recommended that students create a false email account when using GAITs to preserve their privacy. Should you require GAIT use in the classroom, create a general login for all students to use.
- Amazon, Apple, and 12 other major companies that have restricted employees from using ChatGPT"Companies like Amazon and Apple have expressed concerns that AI may put them at-risk of data leaks."
You're smarter than GAITs!
GAITs do not understand or analyze their output. Like their (much less advanced) auto-fill cousins, GAITs just "guess" what word is most likely to come next, regardless of sense or accuracy. They are also known to "hallucinate" or make things up that are completely inaccurate or fictional.
For example, I once asked for recipes that used no more than 1 cup milk, 1/4 cup sugar, and 1/3 cup oil. Even though milk was one of the ingredients, ChatGPT called all the recipes "dairy free."
ChatGPT does not provide accurate citations!
GAITs may provide citations when asked, but few (if any) can generate accurate ones at this time.
- Lawyer apologizes for fake court citations from ChatGPT"But at least six of the submitted cases . . . as research for a brief “appear to be bogus judicial decisions with bogus quotes and bogus internal citations,” said Judge Kevin Castel"
Ethical Concerns
Ethical Academic Use
The first step in ensuring ethical academic use of GAITs in your class is by putting your expectations in the syllabus. There are essentially three options:
- Ban all use.
- Specific and designated use allowed, with appropriate citations.
- Assume and approve use for all assignments, with appropriate citation.
Further information on syllabus statements appears in the How tab.
- Integrating Generative AI into Higher Education: ConsiderationsThere is a duality of AI on many college and university campuses. On the one hand, some higher education officials are eager to adopt AI tools that would assist with student recruitment and enrollment, but on the other hand, many faculty and other institutional staff believe the use of generative AI is a type of cheating or a breach of academic integrity.Footnote7 What is more ethical: guiding the use of AI tools or pretending they do not exist?
Employment Impact
GAITs require humans to train the models, and the companies responsible for doing so often traumatize and exploit those human employees.
OpenAI, the company responsible for ChatGPT, used Kenyan workers making less than $2.00 per hour to train ChatGPT and DALL-E. Many of these employees have developed severe mental disorders, lost their families, and filed lawsuits over their treatment and the materials they were exposed to.
Additionally, companies want to use GAITs to reduce staffing and personnel costs. Many companies have already laid people off in anticipation of an AI being able to take jobs over. One, an eating disorder hotline, had to re-hire humans after the AI gave damaging advice.
- OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less ToxicTime article describing the material reviewed and the working conditions of Kenyans contracted with OpenAI.
- Eating disorder helpline shuts down AI chatbot that gave bad advice"An AI-powered chatbot that replaced employees at an eating disorder hotline has been shut down after it provided harmful advice to people seeking help. "
Environmental Impact
GAITs require significant resources to run. This includes water to cool the servers and many rare metals and minerals. The water use contributes to drought, and the mining for the metals and minerals damages the environment. Additionally, the power use needed to run the GAIT servers is astronomical. These are just a few of the environmental concerns exacerbated by GAIT use.
- A Computer Scientist Breaks Down Generative AI’s Hefty Carbon Footprint"Is generative AI bad for the environment? A computer scientist explains the carbon footprint of ChatGPT and its cousins—and how to reduce it"