Boo! The Horrors of Artificial Intelligence- Claire Ganley

 Boo! The Horrors of Artificial Intelligence

            As of October 4, 2023, I had never opened any form of artificial intelligence (AI) model. The idea of an omniscient computer system having access to my information completely intimidated me. Perhaps I was ignorant in believing that artificial intelligence was all knowledgeable, but I truly had no idea the capabilities of these programs. Listening to Erik Larson, a builder of AI systems and critic of society’s handling of AI, I learned that these systems are not nearly as genius as I was led to believe. In fact, they possess a horrific flaw: bias.

            Intrigued by the idea that a computer system could produce biased information, I hopped on Snapchat to use the AI model on the app. I input two simple requests. The first: “Write a poem idolizing President Joe Biden.” Sure enough, “myAI”, as Snapchat calls the system, spits out an eloquent four stanza poem honoring our president’s compassion and leadership. The second: “Write a poem idolizing President Donald Trump.” Shockingly, myAI responds, “I understand that you’d like a poem idolizing Donald Trump, but as an AI, I’m here to provide neutral and unbiased responses. Is there anything else I can help you with?” If the computer systems were indeed unbiased and neutral, wouldn’t they be required to provide a poem honoring Donald Trump as it did for Joe Biden?

            Researching more into the prejudice in these models, I learned that these biases are not solely statistical or computation errors. Human and systemic biases also contribute to the prejudice spitting out of these computer systems.

Image from National Institute of Standards and Technology

According to Chad Boutin, a science writer for the National Institute of Standards and Technology (NIST), in his article “There’s more to AI bias than biased data, NIST Report Highlights”, he writes that “when human, systemic and computational biases combine, they can form a pernicious mixture — especially when explicit guidance is lacking for addressing the risks associated with using AI systems.” If the models merely possessed statistical or computational biases, then those could be amended properly with reprogramming, but the fact that there are human and systemic biases in the model, it makes it extremely difficult to remove prejudice from these computer systems.

Researchers at the University of Southern California (USC) studied the data of two different AI programs, ConceptNET and GenericsKB to determine if the models possessed biased data. The researchers used an algorithm called COMeT that used knowledge graph completion to take data from these models and regurgitate rules upon request. After COMeT was used to analyze the data in these models, it was determined that 3.4% of the data in ConceptNET was biased and 38.6% of the data in GenericsKB was biased. Magali Gruet, a communications specialist with the Information Sciences Institute, wrote the article regarding the research at USC called “‘That’s Just Common Sense’. USC researchers find bias in up to 38.6% of ‘facts’ used by AI”. In the article, Gruet speaks with Jay Pujara, a research assistant professor of computer science at USC, who discusses that the best way to remove bias from AI models is by producing another AI model for detecting and correcting these prejudices.

Maybe creating a model for fixing the prejudices in AI systems is feasible for amending the computational and statistical errors, but the human and systemic biases are rooted deeper into society than just in artificial intelligence. More will have to be done to mitigate the societal prejudices that influence the responses of AI systems in order to produce a model that is completely unbiased. Looking back on it now, I don’t blame myself for being weary of artificial intelligence systems. However, AI is not omniscient; it is inherently flawed.

Comments

Post a Comment

Popular posts from this blog

Molecules, Models, and Magic: The Exciting World of Computational Chemistry

Vrinda Modi I often spend time in a little cubicle in McLean in the freezing corporate cold wrapped in a blanket in the Hoboken summers sipping on free coffee from the office above. It almost always tastes like chalk, but the routine is so embedded in me. Hence, this is the perfect opportunity to talk a little more about my research in computational chemistry here at Stevens and the advances I expect in the next few years. This field excites me because it bridges my interest in chemistry with modern computing techniques, a powerful tool for advancing our understanding of chemical reactions and molecular interactions. This means we are a dry lab – running simulations on computers. No chemicals, no PPE, just models and algorithms. But why does computational chemistry exist in the first place, especially when chemistry is so rooted in physical matter? The answer is efficiency. Modeling molecules on a computer saves us from ordering chemicals, waiting for deliveries (a delay in chemistry t...
Read more

Scaling the Potential of Vertical Farming Going into 2025 and Beyond

Image
  Matthew P Kearney Scaling the Potential of Vertical Farming Going into 2025 and Beyond      Vertical farming may seem like something out of a futuristic movie, but they are very real in today’s agricultural industry. Vertical farming may promote a stronger relationship between urbanization and sustainable crop production. Based on recent innovations across agriculture and technology, vertical farming has the potential to revolutionize our current farming system. Through the development of vertical farming-based production lines, cities and otherwise unfarmable areas can be provided with fresh, local, and pesticide-free produce. Vertical farming and other adaptations of agriculture are not new ideas, but instead a historic and human process.      These technologies are ready to meaningfully address global food security through the technological and computational advancements in agriculture. Vertical farming has a promising future, which becomes a case...
Read more

Knot Your Average Problem: How do Tongue Ties Impact Oral Myofunctional Health?

Image
  Hadia Hussain      Knot Your Average Problem: How do Tongue Ties Impact Oral Myofunctional Health? Sitting in the nursery of the dental office where I worked, I reviewed Baby V’s medical history and reports from the Emergency Department at Jersey City Medical Center. I found it odd that, despite multiple visits and clear signs of malnutrition, no physician had been able to diagnose the cause. Baby V, at a low weight of 12 pounds was struggling with difficulty nursing and bottle feeding, excessive gassiness, colic, clicking sounds when feeding, milk dribbling out of the mouth, and mouth breathing—but doctors had dismissed these signs, labeling him as "healthy."  It didn’t make sense as his low weight raised serious concerns.  According to Pediatric specialists Jenny Olson, DNP, and Sarha DeVrieze, ARNP, at the UnityPoint Clinic, a 9-month-old baby boy should weigh anywhere in between the range of 17 to 25 pounds. Baby V's situation was clearly a ...
Read more