Thursday, August 3, 2023

Can AI Work for Designers? Or Should We Fear It?

Can AI Work for Designers? Or Should We Fear It?

AI's entry into the graphic design arena has sparked a flurry of questions, like "Will AI art replace artists?" and "Is AI design art really art?" 

To be clear, AI isn't exactly the new kid on the block. Consider Google AI: it's been there since 2016, cleverly curating our video recommendations. So why is it suddenly hogging the limelight? Why the hype, and why now? Essentially, AI has only recently become widely accessible and, frankly, it's become incredibly relevant to designers and artists.

We're going to delve into the nitty-gritty of designing with AI, explore the creation of AI-generated art, attempt to unearth why AI-generated art has been a hot topic in the art world, and ask the question, "Will AI Replace Graphic Designers?"

If you like video content, follow this new video from the Envato Tuts+ YouTube channel:

How Does AI Art Work?

AI is the capability of a computer system to mimic human cognitive functions. It's the ability of a computer to do problem-solving. AI can be developed using several techniques, but machine learning is the one relevant to our modern-day design AI tools. 

Text-to-image AI design tools like Midjourney AI art and Adobe Firefly are two of the most well-known AIs. They can take a string of text and turn it into an image. It's not pulling from an existing image base but generating something from nothing.

Or is it? You have to ask yourself, "How does AI art work?" How does Midjourney know what a "digital painting of a neon cyberpunk girl" would look like if used as a prompt? Machine learning is how these AI graphic design tools know what is what. AIs must be trained on massive amounts of data sets, including images. It uses that data to identify patterns. Eventually, when you say to create "neon cyberpunk girl," it begins to know what that should be based on the patterns in the data it was trained on.

While not all AIs are trained using machine learning, all machine learning is AI.

Machine learning is why some people ask, "Is AI art really art?" After all, it's machines identifying patterns that have to be trained on existing patterns in order to replicate them. 

How AI Graphic Design Tools Can Work for You

As a designer, leveraging the power of AI graphic design tools can significantly boost your productivity and creative process. By integrating AI into your design workflow, you can focus more on the creative aspects of your work, while letting AI handle the repetitive tasks. Here's how you can design with AI.

Asset Generation & AI Visual Design

Asset generation is the first thing that comes to mind with AI design, for obvious reasons. It could be just a mundane object you need, and it's easier to use Photoshop's Generative Fill run on Adobe Firefly to generate an image of a banana. 

AI generated rose using adobe firefly in photoshopAI generated rose using adobe firefly in photoshopAI generated rose using adobe firefly in photoshop
AI generated rose from Adobe Firefly in Photoshop

It could be using Midjourney to create otherwise complex environments that would have been hard to find or make yourself. Using AI image-generation tools for assets will save time, especially if you're using them alongside stock sites.

If you can't find something on a stock site, you can have AI make it, and if AI can't make it, then you can find it on the stock site.

AI generated cyberpunk girls using Midjourney AI generated cyberpunk girls using Midjourney AI generated cyberpunk girls using Midjourney

Inspiration

Current AI graphic design tools can be a great jumping-off point for concepts. While quick concept art and visual references you then use to create something new are a massive appeal for artists, it isn't the only place for AI to be used. 

Khroma

Sites like Khroma train on your chosen color preferences and then generate color combinations based on those results, showing you text, color palettes, and images with those color combinations. Using AI to generate design inspiration and not completed designs is great for people uncomfortable with AI. 

khromakhromakhroma

Adobe Illustrator 

Adobe Illustrator has a tool that will replace the colors in your illustration based on the description you type. Some examples include "neon cyberpunk," "cotton candy," and "pretty pastels."  

I've heard of artists asking chatbots like ChatGPT to request an image concept that the artist then creates, almost like a stand-in for a client. It's one of the more creative ways to design with AI. AI can be an effective art block killer, even for those less inclined to use AI. 

Adobe Illustrator Adobe Illustrator Adobe Illustrator 

Automation of Routine Tasks

Photoshop-based designers are spoilt for choice with an extensive array of automation tools at their disposal, largely thanks to Adobe Sensei. This highly intelligent system has powered Photoshop's auto-select, content fill, and neural filters for years.

Canva for Photo Editing & AI Visual Design

Today, however, web services like Canva are giving traditional design software a run for its money, leveraging generative AI design tools to effortlessly remove and replace objects. As we look ahead, it's clear that web and phone-based editing software are poised for significant growth in capabilities. 

Canva for Photo Editing & AI Visual DesignCanva for Photo Editing & AI Visual DesignCanva for Photo Editing & AI Visual Design

Uizard for Wireframing

Designers are now equipped with incredible tools like AI-powered wireframing and prototyping for rapid project ideation. For instance, Uizard can generate a fully functional, clickable prototype from a mere paper sketch, offering designers the flexibility to decide the level of AI involvement in their creative process.

UIWizaard for Wireframing UIWizaard for Wireframing UIWizaard for Wireframing 

Jasper for Writing Social Media Posts 

Social media and promotional content creation are also getting a facelift with AI tools like Jasper, which generates laser-specific text for posts. While not all designers are fully on board with AI yet, many are excited about these tools and their potential to deliver highly targeted content. 

Jasper for Writing Social Media PostsJasper for Writing Social Media PostsJasper for Writing Social Media Posts

Taskade for Writing To-Do Lists 

I like to use a productivity tool called Taskade. Instead of premade templates that may or may not fit my needs as a content creator, it uses text prompts to create a productivity workflow that fits your exact needs. 

Taskade for Writing To Do Lists Taskade for Writing To Do Lists Taskade for Writing To Do Lists 

Why Is AI Art Bad?

While using AI to generate designs can produce some captivating pieces, it's not all rainbows and butterflies—some argue that the rise of AI-generated art might do more harm than good. Let’s delve into why. 

AI graphic design tools, such as Midjourney AI art, have sparked some hot-button debates. While these tools can generate stunning visuals with a simple prompt, the issue of copyright quickly rears its head, especially when you request art "in the style of" a particular artist.

Taking Vincent van Gogh as an example may seem harmless, but what about current, working artists? Mimicking an artist's style for anything beyond educational studies has always been a faux pas. 

While you can not copyright a style, you can copyright individual works. AI can recreate the work of an artist because it was trained on their copyrighted work.

This brings us to a crucial issue: many of these AI tools, especially AI-generated art tools, might not have secured the necessary permissions for the data they've been trained on. This raises ethical questions, as tech companies could potentially reap massive profits from the work of artists and designers without respecting copyright laws. 

master chief character fan art Created using Midjourney master chief character fan art Created using Midjourney master chief character fan art Created using Midjourney
Created using Midjourney

We also have tools like Adobe Firefly, which proudly claims to train only on images from Adobe Stock. However, their transparency leaves something to be desired. Many users were not given a choice to opt out or even notified past a blurb in their Terms of Service until most of the training was complete. As of July 2023, you can now opt out, but users are opted in by default.

To level the playing field, it might be essential for companies to illustrate how opting in benefits you, the designer, not just them. Emphasizing the value and rewards of participation could go a long way towards resolving these issues.

adobe stockadobe stockadobe stock

Bias and Discrimination

Let's keep in mind that AI, especially within graphic design and image tools, isn't magically self-aware or independent. Instead, it learns from the data we feed it. If we consider an AI that has only been trained on a narrow perspective, its output will reflect that perspective due to the lack of diversity in its training data.

The consequence? Any group underrepresented in the data will be virtually invisible to the AI, thereby portraying a skewed "normal." Hence, both the quality and quantity of data are pivotal in creating an effective AI.

The catch, though, is that these data sets aren't transparent. It's not as if we can sneak a peek at them. The intense competition in the AI field doesn't exactly encourage transparency, but making these data sets transparent is crucial to ensuring the ethical production of design AI tools. 

A Possible Solution With a Content Credentials System

The Content Authenticity Initiative, or CAI, is a collaborative project that aims to set a comprehensive and widely accepted standard for digital content attribution. The CAI claims a strong commitment to ensuring that users can trust and authenticate the digital content they come across.

At its core, the initiative proposes an open standard for tracing the origin and evolution of digital content. By implementing this standard, creators will have a reliable way to declare their work, while consumers will be able to verify the source of the content.

This is done through a Content Credentials system. The Content Credentials system functions like metadata, but with additional information such as the image producer, their social media handles, all edits applied to the image, the editing software used, and the assets used in the image.

The Content Authenticity Initiative or CAIThe Content Authenticity Initiative or CAIThe Content Authenticity Initiative or CAI

How Does the Content Credentials System Work?

As mentioned above, Content Credentials work very similar to metadata, but offer a paper trail of the image's edits and the assets used in editing the image.

What makes Content Credentials even more unique is that individual assets within an image's Content Credentials can have their own credentials. For example, images generated by Adobe Firefly are automatically assigned Content Credentials.

When reviewing the assets section, we can see that the lens flare I created is marked with its own Content Credentials, indicating that it was AI-generated.

During export, you have the option to Publish to Content Credentials Cloud or Attach to File directly. Publishing the Content Credentials to the Content Credentials Cloud makes the credentials more persistent and difficult to remove.

Furthermore, this allows your image, or copies from which the Content Credentials may have been removed, to undergo verification on the CAI's website Verify. If an image is summited to Verify and there's a match, the Content Credentials linked to the original image will be retrieved. 

I exported the below image with Content Credentials to the cloud, took a screenshot of the image, and then ran the screenshot through the Verify site. It found my original image, and brought up the Content Credentials for it. 

If I use my new image as an asset in yet another image, its Content Credentials will still remain intact, just like the Firefly asset did. This creates an Overview tree where the original image, its history, edits, and the history and edits to any assets can all be traced back and connected—as well as any credits to original publishers of assets used along the way.

This comprehensive approach not only reveals the image's creator and their social media, resulting in stronger artist attribution, but also provides insight into the edits made and identifies any use of AI assets.

Will AI Replace Graphic Designers?

Artificial intelligence and graphic design? They're a match made in a designer's heaven! It's truly exciting to see how AI can revolutionize the design process. Just imagine, with a few clicks, you can easily generate stunning color schemes, perfect font combinations, precise wireframes, and visually appealing text layouts. 

Moreover, let's not forget how AI has already enhanced the tools we love and use daily. They've become more accessible, too, thanks to platforms like Canva. 

But why is AI art bad to some people? The ethical implications exist, even if we ignore them or find them inconvenient. Some of the ways AIs are trained are egregious. Some of the ways AIs are marketed are disingenuous. Because of that, I'm uncomfortable using any image-based design AIs in my work and choose not to.

Will AI art replace artists? I don't think so, and while the cat is already out of the bag, and things are being addressed a little late... they are at least being addressed. I firmly believe that there's room for ethical AI in the design world, and we can shape this future together. 

cyberpunk girls generated using Midjourneycyberpunk girls generated using Midjourneycyberpunk girls generated using Midjourney
Made with Midjourney

Check out more tutorials on AI graphic design below! 


No comments:

Post a Comment