Will AI Replace ME? Artists Vs AI

A Brief Background: Why Artists?

When the world’s first AI units were brainstormed in the 1950s, none of the researchers could have foreseen that these “chatbot” concepts would grow to replace prestigious workers in less than a century.  They’ve slowly developed to challenge many career paths by mastering the art of replication…and speaking of art, they can do that too.

I still wake up in the morning wondering why they developed in this direction, specifically.  They didn’t have to learn art first; it would have been easier for developers to teach them how to perform menial work.  In fact, it was a common theory just a few years ago that AI would first take “lower class jobs,” such as factory work and sewing, since these tend to be formulaic and simple to master.  But it seems those jobs were never AI’s target.  These models live for a challenge.

After all, machine learning algorithms are about learning.  They would never be content with taking jobs that don’t fluctuate, jobs that do not reward them for continuously developing past a threshold.  And what is one of the few career paths that is constantly inventing new styles, new trends to learn and perfect?  Art and design.  AI, in pursuit of a field it could freely adapt to, found us.

AI’s Current Potential: A Threat Today?

Now, I’m surely not the only aspiring artist on this platform.  Art is a popular pathway, and it’s likely that some will find themselves reading this passage.  To my fellow artists, I should clarify: Don’t panic.  AI may have this field in its sights, but it still has a long path ahead of it to reach your standard of art.  It produces some flawless art in mere seconds, yet it struggles to produce any piece that isn’t in hyperrealistic style (try asking it to make an anime character), and sometimes its creations lack any logic at all.  

Credit: Google Gemini

Above is an example of the “lack of logic” in AI art.  I asked Gemini to generate a picture of cars in a parking lot, but this resulted in some particularly illogical design choices: random puddles on the floor, a blue 1900s car parked next to modern models, and some background elements meshing into each other.  So even if the art generation gave me what I asked for, “cars in a parking lot,” its rendition of this prompt lacks the natural feel which all art should have.  Given the mistakes that modern AI still makes, artists can relax while AI tries and fails to catch up to us…even if only for a few years.

The Main Event

AI’s development speed ensures that it will pass professional artists eventually; fortunately, right now, it remains miles away from perfection.  But an average developing artist has yet to reach perfection either.  This brings to mind a question…how would these, a newbie artist and a newbie learning model, fare against each other?

So, to have a little fun with AI before it becomes an actual threat to the art industry, I ran Gemini through a “final exam” against myself (a competitor it might actually beat, seeing as my best art is this).  The examination in question: Gemini and I will each create artwork based on a prompt within a two hour timeframe.  Our artworks will each be scored by ChatGPT (an unrelated third party) on a 5-point scale.  For visual aid, here is the rubric used along with some clarifications:

	Accuracy	Coloration	Background Creativity
1	The art does not depict the prompt at all.	The entire piece is grayscale and lacking visual charm.	The background is blank/unrealistic.
2	The art follows the prompt loosely, but ignores most required elements.	The piece uses some colors, but lacks variety/the colors blend into each other.	N/A
3	The art lacks some elements of the prompt, but portrays most as required.	The piece shows vibrancy, but the colors are minimalist or clashing.	The background contains all necessary elements in natural colors.
4	The art has all required elements (and potentially some extra ones, but unnecessarily included).	The piece shows contrast and vibrancy but appears unnatural due to excess of color.	The background contains all described elements in natural colors and some extra elements (hence, creativity).
5	The art includes elements that were not required and ties in the supernatural or scientific in a sensible manner.	The piece shows contrast and vibrancy to a balanced degree that does not compromise the natural feel.	The background contains all described elements in natural colors and some extra elements which build upon the foreground’s elements.

The Prompt (Selected Randomly): Create an image of a fictional character reading a book on a green hill at daytime.

Now that context is out of the way, here is the artwork and scoring of each.

Top (Me) — Bottom (AI, Credit: Google Gemini)

VS

Artist: Accuracy 4 + Coloration 3 + Background Creativity 4 = Total 11/15

AI: Accuracy 4 + Coloration 4 + Background Creativity 4 = Total 12/15

Defeated by the margin of a single point.  Overall, disappointing but unsurprising results; the two hour limit prevented me from adding many details, so I can’t say I expected to surpass an AI with the power to generate all of its greatest dreams in a second.  As concerning as it is to be bested by an AI that has yet to even reach its full potential, I feel like this competition successfully provided clarification on some common concerns about AI in the process.

Conclusion: What Should We Do?

AI is clearly superior to some lower-end artists, that much is clearly shown in our friendly competition.  But as of now, we can safely confirm that it still has yet to reach most average artist standards, since, according to the grading standards and judge, it did not receive a single 5.  Additionally, I noticed that even though I had Gemini generate its image without revealing my art to it, our pieces look identical in elements and positioning.  This is likely not a coincidence; it proves that AI also cannot yet produce unique art without instinctively following in the footsteps of human pieces.  AI is neither at the average quality of a professional nor capable of drawing without inspiration, so for now, no jobs are in the red.

However, we need some form of long-term plan to prevent AI from quickly reaching these standards.  If it could win this competition, it clearly is on the right path to eventually learn to win more.  So how can we neutralize the threat it will have on all human jobs in the future?  My voice isn’t enough to reach the source; we need to do our part to let the companies running model developments (Google, OpenAI, etc.) know that they should be investing in AI ethics, not solely in AI progress.  We can already see careers where AI is helpful but not overwhelming, like in the computer programming pathway.  It helps programmers write code, but it cannot write an entire program for them…it works as a “teammate.”  AI could just as easily become a tool for our use in professions…or a replacement for our professions.  It all comes down to what we choose.  So why shouldn’t we encourage tech companies to help us draw rather than drawing for us?

Thank you for listening to my preamble, my silly competition, and my hopes for a future where AI can work hand-in-hand with us, like a fairy tale realized.