Programming and Politics

No Comment

Machine learning now seems to have a role in every new tech product. Apple has a “more expressive” Siri thanks to “advanced Machine Learning and AI”, Google has a smart camera that uses Machine Learning to take better pictures, and Nvidia recently released a new GPU that contains “optimized deep learning software”.

So what exactly is Machine Learning? In simple terms, Machine Learning is using large amounts of data to find connections between the data and its outputs. This means that Machine Learning can be used to model anything that has a correlation with something else.

Machine Learning’s ability to accurately model almost anything makes it a very powerful tool, but (as always) great power comes with great responsibility. Improvements in Machine Learning models computing power have led to rapidly expanding applications for Machine Learning.

This has created some controversial advances in Machine Learning. For example, a group of researchers at Stanford University recently conducted an experiment where they tested an AI’s ability to distinguish between gay and straight men and women. In the experiment, the AI was given two different pictures taken from a dating website, one of a man or woman who identifies as gay and one of a man or woman who identifies as straight. About 81% of the time in this experiment the AI was able to correctly distinguish between gay and straight men, and approximately 71% of the time it was able to do the same with women.

However, the AI was not quite as accurate in other situations. When it tried to identify 100 men who most likely identify as gay out of a pool of 70 gay men and 930 straight men, it was able to correctly identify 47 men as gay men but missed 23. It also mistakenly identified 53 straight males as gay.

The experiment resulted in a large backlash from LGBTQ organizations such as GLAAD media, who issued a statement claiming that the findings “could serve as weapon to harm both heterosexuals who are inaccurately outed, as well as gay and lesbian people who are in situations where coming out is dangerous.”

This is just one example of a possibly malicious way that AI can be used. As computers and Machine Learning models continue to advance, AI and Machine Learning will increasingly find their way into politically charged and possibly dangerous applications.

In no way does this mean that AI and Machine Learning are dangerous tools that should not be used at all. In fact, I would venture to say that AI and Machine Learning are going to be essential to the future of technology. Even today, Machine Learning is very beneficial humanity and is being used to create self-driving cars, find new ways to detect and prevent viruses and malware, summarize legal documents, prevent money laundering, find new medicines, and recognize credit card fraud.

Clearly, AI and Machine learning are here to stay as they are imperative for a better future, but as their applications widen, it will become increasingly important for researchers and creators to examine the outcomes of their work and act in the best interest of humanity.

Featured Image: https://elearningindustry.com/machine-learning-process-and-scenarios

About the author

Luc Coté is a 16 year old Junior at St. Mark’s School. In his free time Luc enjoys coding, building things, politics, and grilling. Luc is a self-proclaimed grill master, a member of Students for Sustainability, an enthusiastic Peer Tutor, and co-founder of the Young Independents club. Luc hopes to soon start a Coding Club at St. Mark’s as well. Academically, Luc is an involved and passionate student with interests in Computer Science, Math, Physics, Engineering, and History. Luc joined the Parkman Post to combine his enjoyment of politics with his love for STEM.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked (required)

Also in This Issue
Something is wrong. Response takes too long or there is JS error. Press Ctrl+Shift+J or Cmd+Shift+J on a Mac.
Also in this Issue
Instagram
Something is wrong. Response takes too long or there is JS error. Press Ctrl+Shift+J or Cmd+Shift+J on a Mac.