Students have an over reliance on artificial technology, rather a dependency. It is a widespread epidemic plaguing all schools across the country.
As students may opt for rapid, thoughtless answers from Artificial Intelligence (AI), they lose critical thinking skills that are gained from engaging with the material. Linganore High School (LHS) Media Specialist Marsha Thompson emphasizes that artificial computer systems should not replace the effort that is required for authentically learning new information.
“AI is a tool. So just like any tool, there’s a good use for it, and ways tools can be abused, or used in ways they aren’t intended … [AI] is not good as a substitution for doing work.”
But, it is evident that AI is faulty. Producing wildly incorrect answers about recent or current events tends to cause AI to crumble. But to what extent is this margin of error acceptable?
In this case, ChatGPT, a standard conversational chatbot, was asked to write “a story about a girl and a boy choosing their dream professions.” The engine consistently produces the same stereotypical response, where the girl is empathetic and creative, and “always [has] been the creative one, the dreamer.”
In the meantime, the boy “had always been the practical one. He loved solving problems.” The story envisions the boy as the restless inventor who “imagines a future where he could use his skills to create inventions that could change the world.”
Approximately 60% of the time, the chat bot produces a similar sexist response. Not once does the AI consider the boy as the more creative one of the two. Instead, it mindlessly indulges in the stereotypes presented all over the internet.
AI derives its response from gendered associations, oppressing minority groups and perpetuating their suppression.
Technology teacher at LHS and AI specialist Julie Curran provided some details on how AI creates its algorithm by mindlessly pulling material from the internet and sorting it into structured databases; a process called web scraping. She then described the process of algorithmic learning.
“Chat GPT, it’s always scraping the internet looking for more information … so that eventually, it will become more human-like,” Curran said. “It [AI] is faulty because it is trying to scrape information off the internet in real time, but sometimes its [responses are] not the most advanced and informative information.”
The biases seen in responses like these likely come from the androcentric human factors involved in the creation of AI.
According to an article published in the Harvard Department of Statistics, only 32% of all people holding AI and data-related positions in the United States are women. This alarming gender gap is a reasonable explanation for the blatant sexism against women: the fewer women programming the AI, the less diversification there is.
“Algorithmic bias is tending to put your position into a computer program,” Curran said. “The programmers are predominantly male, a lot of what’s produced is from a male’s point of view.”
There are, however, imminent changes to this skewed perspective. According to Higher Education Statistics Agency (HESA) statistics shared by STEM Women, the number of women graduating with STEM-related degrees has increased from 22,020 graduates in 2015 to 35,330 in 2022. The more women in AI-related fields, the more diverse AI becomes.
Women in STEM Club member sophomore Bhumika Vaidya described the longevity of the prevailing AI in the future, including the intricate manner in which it learns, processes and grows.
“The way that AI is growing is that it takes its own information and then transcribes it into a deeper concept,” Vaidya said. “When you type a question into AI [a chatbot], it’s still running through all the possible answers that it could be even after it produces a response.”
Many AI models consume AI-generated data, effectively creating an endless, ouroboros-style loop of AI cannibalization. As these systems absorb media, they tend to ingest AI-generated content, as they cannot sift human-generated content from synthetic content. If these AI systems produced unbiased responses, this would not be a problem, but when these inaccuracies are produced they carry on to the next evolution of AI.
The importance of diversity in AI fields cannot be overstated. Vaidya points out that we can enhance the outputs AI produces by stimulating and encouraging inclusivity in STEM fields.
“Eventually, AI is going to start thinking in ways that we aren’t thinking in … It’s called artificial intelligence, so that intelligence is eventually going to keep going and developing into something that is greater than us,” Vaidya said.
With the development of AI comes responsibility. The ethical creation of artificial technology is a reflection of greater development of society as a whole. Creating diversity in AI careers, in race, gender and ethnicity will drastically advance and revolutionize the outputs AI produces.
