Attestation Of Resistance to Usages of LLM AIs in Academic Settings
Please note that this was originally written for my own personal usage within my experiences in a college setting and within the MNSCU system. Other university and colleges may have different core requirements of students. If you draw inspiration from this, YMMV
Dear professor,
I am choosing not to write this with the academic rigor any typical essay would require. This is not that kind of writing.
If you are reading this there is likelihood that you are assigning coursework that requires and/or encourages the use of an LLM AI to assist or complete. This document is here my declaration and explanation of why I detest such things and why it is my firmest belief that such shortcuts in reasoning have no place in academia.
First a background, as it is also likely that you do not know me well on a personal basis. I am, generally speaking, associated more with arts and creation that technology, though my declared major may have misled you on this. I draw art, I sing, I do voice acting, I create videos, I develop video games, and I write stories. Even this is not a fully exhaustive list of my varied artistic endeavors. I also know the tale of the starving artist. And I have seen this play out time and time again. Hell I’ve lived it for a time. I aim to secure a job that pays me enough money that I can indulge in these artistic pursuits. That is my primary goal in life. And since I find myself somewhat adept with writing code to make funky things happen on computers, I chose software engineering as a solution for acquiring funding.
Hopefully I need to inform you the things that are being done in the technology space with LLM AIs to replace creatives in creative spaces. The use of LLMs immediately after their achievement of baseline passability in the public sphere should be a horrifying concept to any decent person but especially Americans living today who can see the ways our government is choosing to use this technology.
A common reasoning for the “value” of AI generated arts is that it “democratizes” the artistic process. I call absolute bullshit on that. Art has never been a thing that requires much of anything. You can take a bus to the local library, or walk if it is close enough, and use a freely available computer to use a freely available word processing application to put a series of keystrokes together to form a series of sentences and phrases and altogether it creates art. The claim of using LLM AI assists for creating of art due to lack of skill only furthers the demonstrably harmful fixed mindset. My entire childhood teachers were fighting tooth and nail to help students learn growth and how to focus on that growth. I grew up watching Bob Ross paint before the morning cartoons. As he often said, “Anybody can paint.”
Another issue I have with these LLM AIs is the generation of fluff. Why do we have word requirements in essays? Why do we write emails? Why do we communicate with other flesh and blood human beings? To me, the usage of an LLM to “create” these kinds of things says to me: “I couldn’t care a single drop less about this, but I expect you to care about it enough to ingest it in its entirety.” What is the value added by one person using an AI to expand their bullet point list of information into a multi-paragraph long email if the other person is then expected to use an AI to summarize that. A double-ended non-deterministic filter is being applied on the information that one person is trying to express. You wouldn’t want to dilute your message with fluff or risk a misunderstanding in effective communication. The usage of LLM AIs in this space only serves to degrade the quality of communications while selling the solution to a problem that it creates.
So to continue the thread of monetary affects the surge of LLM “creation” has, it is worth looking at economics as well. Even the most basic plan (anything above the free tier) of ChatGPT has been $20 USD per month, which runs $240 per year not accounting for taxes or potential changes in pricing. Now oppose that to the yearly gross income of someone working full time while earning the federal minimum wage. They would earn $7.25 per hour, reaching $15,080 per year ( $1,256.67 monthly average). That means that a worker earning the absolute minimum that a company can pay them is just over a 500% markup in price. Another notable element that I doubt I need remind you is that we exist in a free market capitalist economy. Any entity within this economy is able to set prices as they see fit proportional to demand and its elasticity. In the United States, having effective health care all but requires employment with a decent company. Likewise the typical cost of living in any given area typically exceeds that of a minimum wage worker. Simply put, the typical American cannot afford to live here. And they can’t afford to leave. And now we introduce a technology that gives companies a steep discount on employing these workers. LLMs do not require a benefits package. They don’t require breaks, or food. All they require is all the water on our planet and every single joule we can extract from our dying planet’s remaining natural resources. When I look and see what is happening in the world with respect to AI, I do not feel hope. I feel fear and abject horror and the atrocities that are completely avoidable. But the demand of our market is that any given company raises their profits year by year. Where is this infinite growth going to come from? Is there a spare planet for the coal and oil to come from? An exoplanet that we can harvest hydrogen and oxygen from? Are those methods even reasonably adequate to afford us this reckless cost to our resources. A single query of ChatGPT is expected to consume roughly 0.001 to 0.01 kWh. To put that into proper context, a web search using google consumes roughly 0.0003 kWh. Assuming purely linear energy consumption of web searches, one could perform 3 to 33 google web searches before consuming the same energy as asking ChatGPT..
Time and time again throughout my life I have watched “once in a lifetime” economic events. My family has experienced a number of hardships just to get to today. I can only even afford to pay for this schooling due to a payment plan my company that I work for is helping me do. And that is an entire other discussion that I have equally strong thoughts about. But to avoid getting too far off track, please understand that this choice is not a victimless action. Ambivalence and centrism only serve the dominant force.
Now to discuss the academic side of things. Personally, I would state that the primary value of attending formal higher education (post-secondary education, colleges, universities, I’m going to oversimplify that a bit) is to gain a particular set of skills that can be valuable for an individual’s life and career. Notably:
- Experience from and with professionals in the student’s field of interest
- Gain valuable critical thinking skills that enable one to experience and interrogate the world around them in professional and private settings
- Develop oneself into a member of a productive society, contributing to the well-being of themselves, their loved ones, their community, and their country.
It may be important to note the core goal areas required of any graduate in the Minnesota State Colleges & Universities system:
- Communication
- Critical Thinking
- Natural Sciences
- Mathematical / Logical Reasoning
- History and the Social and Behavioral Sciences
- Humanities and the Fine Arts
- Human Diversity, Race, Power, and Justice in the United States
- Global Perspective
- Ethical and Civic Responsibility
- People and the Environment
It is my firm belief that any person willing to contend with the expected goal areas of our educational system would be able to understand how harmful the rampant use of LLM AIs is. Any one of these goal areas should provide enough reason to avoid the use of LLMs.
We are not “living in the future”, “democratizing art”, or “saving time and effort”. We are killing our planet and ourselves while laughing at shitty pictures of generated clones of existing artwork.
Note once again that while this is not a formal essay or academic paper, I felt the urge to provide some resources that I used for my reasoning and estimations.
- https://www.npr.org/2024/10/18/nx-s1-5153741/ai-images-hurricanes-disasters-propaganda
- https://www.businessenergyuk.com/knowledge-hub/chatgpt-energy-consumption-visualized/
- https://dev.to/nilanth/how-much-energy-does-chatgpt-use-per-prompt-a-look-at-its-hidden-environmental-costs-2j3a
- https://kanoppi.co/search-engines-vs-ai-energy-consumption-compared/
- https://medium.com/@sohail_saifi/the-real-environmental-cost-of-every-google-search-with-actual-calculations-6b72e2dd7f3c
- https://www.minnstate.edu/admissions/transfer/mntc.html