Mad about Science: Artificial intelligence, redux

By Brenden Bobby
Reader Columnist

A few years have passed since we’ve last explored artificial intelligence. The last time I wrote an article dedicated to the subject was in 2018, and the scope and reach of AI has changed considerably in that time.

Artificial intelligence in 2018 was essentially contained to advertising algorithms — limited uses of predictive text and video recommendations on streaming platforms. Many of those purposes have become greatly refined in the past six years, while a huge number of new applications have blossomed since the dark age of COVID.

Let’s begin with generative AI, a tool that I, as a creative, marvel at with both disdain and bewonderment. The idea that a machine will be able to replace artists is a laughable concept to virtually any creative spirit and the ethics of using generative AI to shortcut creative purposes is a deep and convoluted topic and one we won’t dive too far into here. However, generative AI does serve some useful purposes for elevating everyone without seeking to damage creatives exclusively.

Image created using generative AI.

Generative AI is frequently used during the process of grantwriting and drafting mundane emails suited to particular audiences. It’s also useful for marketing purposes and other nuts and bolts topics that aren’t necessarily fun to write about and don’t require much “soul” to convey a point. It is not a surrogate for journalism, fiction writing or anything that’s trying to convey the soul or passion of the writer as the computer is essentially just reproducing data inputs in an attempt to replicate the way a human communicates.

Here is the problem with generative AI: It requires data inputs in order to function. The more data it is fed, the more biased it becomes — unless certain parameters are set by humans. This means that, at some point, it’s plausible to believe that generative AI may create some form of feedback loop where it’s essentially feeding itself input data from many users like a big digital ouroboros. 

In many ways, generative AI can be used to elevate substandard writers to a more professional or refined level — this is a process of raising the average by elevating the lowest levels of a curve, but it will ultimately plateau in the process. It may also have the opposite effect by lowering the peaks of the quality curve and creating more demand for mediocre communication and “dumbing down” interest and taste in writing.

Generative AI has even found uses in crafting curriculums for schools and homeschool students while adhering to specific state standards when fed those parameters by the user. AI assistants can be paired in with these generated curriculums to aid the student in their educational journey, without simply giving them all the answers. Schoolai is a great example of this and is particularly useful for learning how to code independently.

A disturbing trend has been observed at a very large scale through recommendation systems, particularly in streaming services like YouTube. The algorithm behind YouTube is influenced both by user trends and inputs from Google, with the intention of selling more ads. Shorter videos means more frequent ads, which means more ad revenue. Shorter videos are also more digestible to the viewer as one can watch five little videos during a rest break or a brief reign atop the porcelain throne. More viewers watching shorter videos skews the algorithm into recommending shorter videos to everyone, while Google turning the dial to favor this will push it even further.

This was directly observed to influence the outside world and society at large when algorithmic junk was fed to young viewers during key developmental periods in their growth. In 2017, Youtube’s Kids app was plagued by cheap nonsense-drivel videos with assets stolen from Marvel, Disney and Nickelodeon designed to be played on loop to generate cheap ad revenue. Young children left to the digital babysitter were observed to have spent hours consuming this content, feeding the algorithm into more and more disturbing nonsense videos they would then parrot to their parents and peers.

Artificial intelligence presents a number of ethical dilemmas, but it is not some omnipotent being or digital overlord with complete and total sway over our every decision. It is ultimately controlled by people with the intention of making money. It is important to be cognizant of the reasoning of why when utilizing AI. Are you utilizing AI to fast-track mundane parts of your life or a job that you already know how to do, but don’t want to waste hours accomplishing in order to get to the really meaningful stuff? Are you using it to simply bypass things you don’t want to do, but may not necessarily know how to do?

AI-generated digital art is a hot-button issue right now. As an artist about to earn the ire of the entirety of Bonner County, I believe it has a place in our repertoire as artists. Certainly, it’s wildly unethical that AI companies have been stealing artwork to feed the machine without crediting or paying the artists their fair share while reaping massive profits for themselves. However, AI-generated art can be a fantastic source of reference material when trying to mesh very specific color palettes or trying to find those perfect reference images for something you’re trying to paint or draw. 

It can also be used as a source of inspiration, to help push you to create the piece you really want to create by using that weird, nightmarish landscape it generated as a spark of inspiration or reference without directly copying it.

Like it or not, AI has permeated every aspect of our daily lives. That being said, it has never been more important to ask the question: ”Why?” at every available opportunity. 

Can you identify every place AI has influence in your life? You might be surprised to find out just how pervasive it is.

Stay curious, 7B.