Exploring the Risks of AI Art

It was requested that I give a breakdown of my thoughts on AI art. Here is a breakdown of the four main aspects that I most commonly hear about associated with AI art and thoughts on them.

The aspects are: job security, intellectual property (IP), resource use, and the terminator problem. This article will be part one looking at Job security and intellectual property, next week will cover the other two.

Job Security

One concern raised around AI is that it will replace jobs. Some artists have raised the concern that computers are capable of duplicating their work and risk making significant numbers of individuals in their fields obsolete.

There are different arguments surrounding if the AI generated content is inferior, superior, an IP infringement, or other things, but fundamentally the primary issue here with this aspect is job security.

Two main factors inform this particular issue; I think of these as market demand and critical infrastructure, but people may have other names for these things.

Every good and service is subject to market demand. If AI products are inferior, but the consumer would rather purchase them, then the consumer has ‘voted with their wallet’ and has concluded that the inferior AI products are actually superior.

If the above is not correct because the consumer cannot be trusted to vote with their wallet, then an alternative should be suggested. A precedent for restricting purchases already exists. The state already restricts certain goods and services like illicit drugs, murder-for-hire and weapons that are a significant threat to public safety.

Given how extreme a good or service usually needs to be to justify this sort of restriction, AI generated content would likely not qualify, but perhaps a novel argument could be made.

Critical infrastructure is an interesting angle. A country will determine what its critical infrastructure is and will apply special protections for them. These things are generally considered core to the country’s ability to function. Artists could make a sort of cultural argument and argue that human-sourced art represents a sort of cultural value that is critical to the country’s existence. 

Given that education is not currently a critical infrastructure, this will be difficult to justify. Critical infrastructures have significant safeguards from disruption. In many countries around the world, teachers and other educational staff are not an essential service. They have a right to strike, which is the ability to disrupt the infrastructure.

Staff that ensure the functioning of critical infrastructure will usually be classified as an essential service and will gain certain privileges. This is because allowing them the right to strike can represent an intolerable threat to public safety or national security

Whichever way this is argued, artists have to justify why they are ‘special’. If the market decides that artists are inefficient, then the ask is for the government to support the industry. This means redistributing income from other people to artists in one way or another.

The artists need to have an answer for the question posed by the average voter: “why do you deserve to be paid by my income?”. If artists are going to be supported then someone has to pay the taxes to support the artist that cannot survive in the free market.

Personally, I think making a guaranteed annual income (GAI) or universal basic income (UBI) sort of argument is easier to make than supporting a specific industry.

Ireland introduced an arts sector targeted GAI. The pilot program gave 2000 randomly selected eligible artists €325 per week ($1,300–$1,500 monthly) for three-years (2022 – 2025). In 2026 the program was made permanent. Due to the GAI program being targeted, these artists were effectively ‘special’.

Most of the other people that lost their jobs via technological advancements throughout history had similar criticisms, and if everyone gets paid (as opposed to only artists) then one does not have to justify why they are ‘special’.

Intellectual Property

Here are three issues with intellectual property and generative AI.

The first issue, some have argued that the way some generative AI technically works involved a process that is, by-definition, not stealing work. This is model-dependent, but apparently some of these models have been designed in such a way that they don’t even interact with protected material. For the following 2^nd and 3^rd issues, we will assume that AI is breaching intellectual property rights, because if the first applies then the rest doesn’t really matter.

The second issue, the situation is legally tricky. Sometimes the person using the AI is using a free program and realistically doesn’t have the assets to be worth suing. It costs more in legal fees to sue them than the plaintiff can expect to collect. 

Other times, the artist has very few resources and may be attempting to sue a tech giant because of how their model behaves. The artist doesn’t have the financial resources to have a reasonable chance to practically exercise their intellectual property rights.

The third issue, some players in the market are not bound by intellectual property and don’t care. As an example, assume the United States decides to halt all AI research and China decides to accelerate it. China could eventually get to the point where they are able to produce so much entertainment media so quickly that the United States consumer will prefer the Chinese product. 

China does not recognize international intellectual property, they don’t care. In this scenario the United States has limited its own market and has effectively ceded its dominant position in media to China. If the decision becomes one between protecting workers and a threat to the sector as a whole, the workers will likely lose.

Next week will look at the other two aspects: resource use, and the terminator problem.

Learn more and watch the full video here.

Vezina is the CEO of Prepared Canada Corp. and is the author of Continuity 101.

He can be reached at info@prepared.ca.