It's now fairly clear that ai is going to eliminate a lot of jobs in the service sector and some creative roles are not going to be impervious to this disruption. Exactly which roles and skills will be made redundant by ai remains to be seen. To search for an answer let's take a look at what it can and cant do as of now.

Ai generators are great at rendering 3d scenes, characters and objects from chatbot prompts. The asset sourcing the ai uses as reference is probably unethical, but we are not going to be able to regulate it anytime soon. The lighting, texture and character and landscape modeling are so good now that if I were studying to be a concept artist at this point I would change my major.

Ai enhanced design and layout programs can also generate iterations changing backgrounds out and providing quick variations on color scheme, backgrounds, font pairings etc. The number of entry level design and production jobs will probably be cut substantially.

This layout iteration ability will also effect the social media content manager role. This role however, includes a diverse set of tasks, including copywriting(which ai can do better than most humans) and it remains to be seen if the ai disruption will be a net positive or negative.

What about identity and brand design?

Here is where I see some hope for traditional graphic designers. Vector art generators do exist but they need more input that text from a chatbot.

Kittl is a specialist ai vector generating app that works by using a clip art library as an secondary prompt to filter styles for the text prompt. Small freelance graphic design jobs, where someone just needs a quick logo or simple layout to get off the ground, or fill a social media post are being replaced currently by this. However, when a business needs something more original and specific the traditional designer has the upper hand.

This S was created from evolution of the text prompt "create an S with spiral serifs" in nightcafe.

Most ai art generators couldn't create a simple vector like this from a text prompt.

Ai does have the ability to trace over an existing raster image and create an accurate vector. This is great for creating vectors from sketches or Shepherd Fairey type images, but this tech has been around for a while in vector drawing apps. Below example created in vectorizer.

Even simple geometric line work is difficult for an Ai generator when prompting from a chatbot or clip art library.

The two examples below illustrate the contrast between traditional vector art and vector Ai from a text prompt and clip art filter. On the left is a a simple spiral motif rendered from a sketch using Adobe capture and tidied up in Illustrator. On the right is the closest approximation I could get kittl to output.

The pilot / ai co-pilot workflow works currently, especially with vectors and layouts.

The above methods are not going to replace too many design jobs. Even using the vectorizer app to trace a sketch isn't an efficient work flow for a designer with access to professional tools-though it could work well for non-designers.

Hybrid workflows and Image based prompts. Apps like Adobe Firefly will take a sketch prompts for vectors, or Flair that uses image prompts for variation... will potentially be the standard workflow going forward. In these workflows ai handles the production work but the designer provides the big idea.



Adobe firefly is pioneering the sketch to vector prompt as shown above, such image prompting will both help and hurt the designer.

The big concern with these image based prompts would be unethical people using the work of others as iteration prompts to create finished work without human input, technically not plagiarism, under the current regulations, but still highly unethical. This could potentially be the biggest killer of design jobs.

My advice to a design student would be to focus on heavily vector based work, and also look at the work of the deconstructivists, from the late 90s who were reacting to the industry disruption at that time.

See more examples of some of my recent vector work that AI would have a hard time creating.