March 15, 2023

Digital

The AI Controversy - Chat GPT, Midjourney

Mass Hysteria Across Creatives

Artificial Intelligence has swept the world off it's feet; Chat GPT, Midjourney, DALL-E, and others have introduced new ways of creating content. While most of the news outlets praised the advancements, it struck one community hard: creatives. We are all going to lose our jobs... well, not quite.

AI Generated Art.

First, it is essential to define AI-generated art. AI-generated art is artwork that is created by an algorithm or machine learning program, like MidJourney, DALL-E, and Stable Diffusion. These programs are typically trained on a dataset of existing artwork, which allows them to generate new pieces that are similar in style to the original works, as well as understand what different subjects are. The resulting art can take various forms, from paintings, drawings, to photorealistic artwork.

The Pros:

AI-generated art is accessible. AI algorithms can create artwork quickly with just a short text prompt, making it more accessible to people who may not have the resources to create art otherwise. I can create a logo, but I lack the ability to create abstract street art. If I input "abstract red and black street art", within a minute, it's created a full blown artwork while I'm still opening photoshop. If I were to go more in detail "Minneapolis Horizon in the style of Graffiti Street Art, red and black", it outputs the concept out fairly quickly. (See Below)

Image
MidJourney's Response to "Abstract Red and Black Street Art"
Image
MidJourney's Response to "Minneapolis Horizon in the style of Graffiti Street Art, red and black"

One of the potential benefits of AI-generated art is its ability to push the boundaries of creativity. Would I have been able to create the artwork above? If you gave me a couple of hours, I could make something with a similar aesthetic and probably more flushed out, however, without a proof of concept, I would have no idea where to begin. AI algorithms can create artwork that is unique and groundbreaking, incorporating elements that artists may not have pictured. In this way, AI-generated art can be seen as a tool to inspire human artists and help them develop new ideas and techniques.

Despite these potential benefits, there are also several arguments against AI-generated art.

The Cons:

AI-generated art is not truly creative. While the algorithms used to create the art may be innovative, they are still programmed to follow specific rules and guidelines. Critics argue that this makes the resulting artwork predictable and lacking in spontaneity, leading to a sameness across all AI-generated art pieces. The software is limited to your creative input. If you input "cat in a hat", you're going to get a basic image output. it still requires people to direct the creative direction, and even then, it lacks the emotions we want to feel from the artwork.

Image
Midjourney's Responce for "Cat In A Hat"
Image
MidJourney's Response to "Russian Blue cat wearing a fedora staring into the sunset on the Eiffel tower"

Another controversial aspect of AI-generated art is the issue of authorship. Who should be credited as the creator of AI-generated art? Although the user is giving creative input to the software to create the image, it's taken inspiration from it's database, many from artists who did not consent to being added, and spitting out images that could potentially fall under plagiarism or in violation of copyright.

Take for example a recent case: Zarya of The Dawn. This is a comic book created by Kris Kashtanova, from the ground up - except the artwork. The text is hers, the concept, even the input towards Midjourney to create cohesiveness within the designs instead of looking like different artists and renders; all her work. However, because the images were created by AI, thus taking inspiration from the database, the images pertaining to the comic book had their copyright protection revoked.

The Art of Diffusion

This is a message to creatives out there. Is your art original?

Sure, you're not directly stealing images as inspiration and claiming them as your own, however, we are doing what AI does: sampling data. Everything is a Remix as Kirby Ferguson puts it. Whenever we build a logo for a client, or a website, we create a "vision artboard"; a way for us to visualize different assets from the industry to build a vision. The final project isn't a direct copy, however, I looked at data to create a design.

Everything the AI is doing is legal, but since it feels like it may be coming for our jobs, there's been an outcry of "Support Local Artists" and petitions signed to remove data from the software's database, and here are my thought's behind it:

We're not going to lose our jobs.

The initial training for the software should've been an opt-in service, and trained with public domain art. By training through active communities like DeviantArt, they are going against the wishes of many creators to use their artwork and potentially being susceptible to copyright issues.

That being said, if someone attempts to create a logo on Midjourney, good luck! The files that are spit out are blurry, and vectorizing it is going to be a pretty fee. With AI Algorithms, to create the exact artwork you are looking for, it still needs a creative's perspective to make the minute changes you are looking for; something AI can't handle.

For those doing commission artwork, the value isn't only behind your artistic style, but if not the emotional impact. They want YOU to make it, not AI, not someone with a similar aesthetic.

AI is a tool for us. As mentioned previously, it has the potential to showcase us inspirations. I know I personally like having a rough idea of what I'm trying to create to pursue the vision. My office wrap (courtesy of Sun Control of Minnesota) started out as a Midjourney idea, tossed into photoshop, cleaned up, and added elements to increase the sharpness, and better convey the feeling I wanted. If a client doesn't have an idea for a logo, we can showcase them AI generated designs for inspiration to get a sense of what style they are looking for.

Office Wall Wrap

Artificial Intelligence vs Artificial General Intelligence

Artificial Intelligence is hard to describe, but the general consensus is programing computers to do things that traditionally require human intelligence.

The Terminator Poster 24 x 36in

Artificial General Intelligence (AGI) is what we see in Terminator and Ex Machina. The main distinction between AI and AGI is the machine being able to comprehend the data and learn any task that a human could complete. Taking for example NotionAI, using it's current dataset, it could create content per your request, but because it lacks the ability to understand the dataset and create it's own hypothesis, it's regurgitating the data it has. If you ask for something outside it's capabilities, instead of attempting to learn, it will communicate with you one of two options:

  1. The Software will communicate with you that it's unable to complete the request due to lacking data, or having that dataset locked by the administration
  2. It will grab dataset from a similar conversation and make up it's own facts. The way Linus and Luke from Linus Tech Tips put it: the software will be confidentally wrong.

Professionals in the Artificial Intelligence field are split down the middle on the future of Artificial General Intelligence; some estimating we are going to see it soon, or that Artificial General Intelligence will never exist.

The reason I bring up AIG, is that ChatGPT, NotionAI, Midjourney, any Artificial Intelligence out there that has been hyped up by News Outlets and Youtube are not Artificial General Intelligence. All these softwares have extensive datasets to create their outputs, and have an algorithm programed to make them sound human, but they cannot create their own ideas.

Creating Written Content with ChatGPT

Supercharging Your Product Design Workflow with Chat GPT and Figma | by  Judith Lopez | Mar, 2023 | UX Planet

Don't get me wrong, this is one of my favorite tools, but it's important to treat it as inspirational. It's unable to replicate your company voice for social media, it's unable to produce high quality blogs since it follows the same five paragraph structure taught to us in middle school. Feel free to use the content, just remember with the development of AI also comes with AI detectors, which could hurt your SEO in the long run.

What it can do, is give you a rough outline for a blog where you can add your own content and experience. It can give you some blog topics that would be interesting to discuss, that also target the right audience.

Remember that if it's a topic that isn't well documented in it's database, it's going to be producing content that cannot be trusted. It's compiling data and summarizing it with an algorithm that makes it appear like a person wrote it.

it can save you time with outlines, but it cannot replace a copywriter / marketer.

The True Market for ChatGPT

Again creatives, ChatGPT isn't going after your jobs, it's a tool we can use, but it's not just for us.

Microsoft has invested 10 billion into ChatGPT to be able to use the software directly in Edge and paired with Bing.

Why? The goal with the partnership is to make web searching easier. Instead of having to dive into the search engine at multiple pages, the search engine can now spit out an answer to what you're looking for. If you're reading through a news article, but don't have the time, AI can now summarize it for you using the content from that webpage. While Microsoft has ChatGPT, Google is working on Bard to achieve this same effect.

So what am I getting at? Instead of assuming the worst with AI and boycotting it, we should learn the tool, just like we learned photoshop, WordPress, and social media. Adapt to the circumstances and use it to your advantage to save on man hours and improve your turnaround times.

Related post