The Internet: Social Media & Algorithms

The Internet:

Social Media & AI

The internet is now decades old, but the popularity of social media began to soar in the mid-2000s. Artificial Intelligence is a newer internet development, first released to the public when ChatGPT debuted in late 2022.

The internet has had ample positive impact. It allows people to connect with others regardless of location, reuniting friends and family and allowing people to expand their social and professional circles. The internet also allows businesses and other institutions to operate quickly from different places around the world. It allows people to start their own small businesses. And it allows people to access information instantly, from home and often for free, in ways that were not previously possible.

As the internet grew in popularity in the late 20th and early 21st centuries, many business and political leaders became "internet optimists." This meant that they saw the positive ways the internet could advance the organizations they led without creating norms, practices, or laws to forestall its potential negative impact. Everyday people, too, became internet optimists, putting their interests, their work, and often their personal data online. Today, people born in the 2000s may have an internet footprint as old as they are.

The readings in this section offer definitions of two key internet-based items, algorithms and artificial intelligence. Algorithms inform our social media feed. If we get all of our news from social media, much of what we learn about current events is based on an algorithm. Generative Artificial Intelligence programs like ChatGPT are also algorithm-based. Generative AI creates new text based on user prompts. The algorithm uses documents fed into its system to answer users' questions. GAI is a long-form version of the same type of program that suggests words to finish a text message based on the first few letters input, like when you type the letters "qu" and your phone provides the option of clicking one button to finish the word "quickly."

Unfortunately, the "endless scroll" of social media creates incentive for extremism rather than credibility. It's human nature to be drawn to an extreme headline over a modest one. Take these examples of headlines describing the same events:

I'd like to think the former headline is more accurate, but the latter is undoubtedly more appealing. Words like "deranged," "mad," and "insane" stand out, making scrollers more likely to click the extreme headline than the modest headline. Through this incentive structure and the way we sort ourselves into friendship/affinity groups based on our interests, social media extremism creates a culture with more extremism. This social environment can even lead users to sever relationships with friends and family who don't follow the same news and develop the same beliefs.

While social media fosters extremism, programs like ChatGPT are trained to create a different kind of emotional appeal. AI responds to a reader's language choices and often provides users compliments in comforting tones. The reassuring language ("that's a great idea!") is designed to keep users on the program, spending more time on it for all kinds of information or advice. While social media sorts people into groups, AI is designed for a one-on-one relationship with the user.

Both social media's algorithm of extremism and AI's comfort-based algorithm create problems. AI software is only as good as the information fed to it, and programmers can skew algorithmic responses fed to users as they see fit. The same is true with social media feeds.

Over time, people who use AI too much will have less capacity for critical thinking. Even if the AI software is programmed for objectivity and accuracy, a person who uses it too often isn't using their brain, so the brain becomes less capable of solving the problems AI is used to solve. Think of the brain like a muscle. Training it repeatedly makes it stronger.

AI has the potential to hallucinate and provide inaccurate information as a result. I was a victim of an AI hallucination recently, but thankfully, it was about a relatively minor issue. I used Google's AI software to find out how many bands were playing a show I had tickets to. It told me there were 3 bands playing that night. I nearly missed the headliner because the program hallucinated and combined 3 nights into one in its response to me. The other acts actually played on different nights of that week. This was a hallucination because it's quite doubtful Google programmed its software to trick me into missing a concert.

There are major concerns about AI programmers with ill intent. If programmed with a particular goal in mind, AI can intentionally provide false information, about all kinds of things, not just musicians' touring schedules. AI developers can create an algorithm that distorts historical facts or scientific truths, for example. And we've mostly focused on text-based AI. AI graphics and videos can create an even more powerful doubt in viewers. Some AI video and audio is made to misinform. Even the types of AI audio/video that doesn't intend to misinform listeners and viewers contributes to an atmosphere of distrust. People now doubt the credibility of real audio and video evidence because it could be AI.

Some users see text AI as omniscient, i.e. knowing everything. Users who view AI this way are less likely to think for themselves and through over-use of AI have less ability to think critically. Meanwhile, conspiracy theories are a primary staple of extremist social media feeds. So, as noted at the beginning of this piece, the early "internet optimists" correctly noted major benefits in fostering the growth of the internet, but the connectivity the internet brings has also featured prominently in the rise of conspiratorial thinking.

The articles linked in this section focus on:

Defining an algorithm (Deny)
Defining problems with social media influencer culture (Ekinci)
Discussion of Artificial Intelligence's impact on human thought and ability (Abrams)
Defining parasocial relationships and understanding if and when they are a problem (Ma)
Discussion of the ways social media algorithms influence relationships (Bishop)
Study results highlighting the problems with overreliance on Artificial Intelligence (Conner-Simons)

Sources:

Local Professor Develops Earnest Website to Help People Counter Conspiratorial Thinking

Deranged Mad Professor Makes Insane Anti-Conspiracy Website Because He Thinks He's God Or Something

Contact

Jay Trucker, Ph.D.

Email

jtrucker@ccbcmd.edu