I don’t know if you’ve noticed, but 2023 might just be the year SkyNet takes over, if I may take a reference from the Terminator movies.

Artificial Intelligence has gone from a buzzword / niche-technology to a quickly maturing industry with platforms for art, copywriting, presentation creation, logo design, and many others…all which present lofty promises, endless creative possibilities, and a dash of existential dread for anyone in creative design.

So, what does that mean to your business?
Do you still need a creative agency?

Here at BOLTGROUP we are quite excited about AI; we have closely followed the various technologies, tested them out, and begun to apply them to our Product Design, Brand Identity, and Visual Brand Language workflows. Overall, we see AI technologies as being powerful tools to speed up our process, but they do not replace people. At least not yet.

AI tools provide incredible opportunities to create interesting creative content more quickly than at any point in human history. The human imagination is endless, but our time and technical skills can limit us at times. AI offers us the opportunity to unleash our creative vision by learning a few strategic prompts while wrapping our heads around both the potential benefits and the pitfalls of the technology.

The major tools that AI platforms lack are strategic vision and an overall process of how to create an innovative product or a meaningful brand, and how the two relate. They also lack empathy and an editorial eye to know the difference between good creative expression and something that lives deep in the uncanny valley.

“It’s irresponsible to use this technology without understanding that it tends to hallucinate and lie…”– Pau Garcia, a thought leader in integrating AI into the design process

My best comparison is that these AI technologies are the equivalent of talented but extremely quirky new team members that can produce interesting but albeit derivative content.

• They generate highly realistic images insanely fast.
• You are not quite sure where they are getting their ideas. The outputs range from what might be copies of source images, to inspiring art, to bizarre and creepy images.
• They tend to only save their work in annoying formats and don’t follow editing directions well.
• You are probably going to need to redo the work before you show it to clients or before it is ready to release (in the case of UI).
• Sometimes it seems like it’s as much work cleaning up their output as it is to just do it yourself right the first time.
• They are inexpensive and come up with ideas you might not have thought of yourself.
• They are absolutely improving on a daily basis.

In this blog post, we will cover some key creative touchpoints as I run through an imaginary brand and product release. I’ll be using AI powered creative writing and visualization tools / platforms like ChatGPT and MidJourney, which use natural language processing (NLP) and machine learning algorithms to generate written content and visual representations based on user input.

Currently, AI cannot complete all of the prescriptive steps of a typical Industrial Design process—like testing, refining, or implementing—but is very good at others, like generating ideas. Here is an example of some key steps where AI tools can be leveraged by human designers.

Let’s note that Chat GPT can be pretty good at research, though you need to be careful to double-check anything it generates for accuracy. I am focused on the creative process, so I am going to keep this exercise generic and start with brainwriting, and not go through too many iterations to show both the power and the limitations of the software.

• Chat GPT (https://chat.openai.com/chat)

One of the best things about Chat GPT is how open-ended it is. Since I started writing this blog, GPT4 was released, offering even more nuance and power. If you want it to write ideas in German it can. If you get deep enough in the weeds you can give its samples of your style and it will learn to write like you. I find it particularly helpful in creating or fleshing out outlines. With writing, the most difficult part is often starting and Chat GPT will give you a starting point instantly. It is worth finding some “cheat sheets” to explore the possibilities and tricks like this one https://quickref.me/chatgpt.html, but I would suggest looking around because this technology is changing very quickly.

Input to Chat GPT
“Generate three ideas for a sustainable new shoe brand.”

Output from Chat GPT

1. “EcoSole”—A sustainable shoe brand that uses recycled materials to create unique designs. The shoes would be made from recycled plastic, rubber, and other materials that would normally end up in landfills. EcoSole would also utilize eco-friendly production techniques, such as using solar-powered factories and minimizing water usage. Each pair of shoes would come with a tag that lists the percentage of recycled materials used in its production.
2. “GreenStep”—A sustainable shoe brand that focuses on using organic materials such as bamboo, hemp, and organic cotton in its designs. GreenStep shoes would also be made using sustainable production techniques, such as minimizing waste and using renewable energy sources. The shoes would come in a variety of styles, from sneakers to sandals, and would be designed for both men and women.
3. “CarbonFootprint”—A sustainable shoe brand that aims to offset its carbon footprint by using renewable energy sources, such as wind and solar power, in its production. The shoes would be made from sustainable materials, such as organic cotton and recycled rubber, and would be designed to be durable and long-lasting. Each pair of shoes would also come with a carbon offset certificate, indicating how much carbon was offset during the production process.

Analysis
These are excellent starting points with ideas for the brand names, brand story, materials, brand communication ideas, and what makes the idea unique and sustainable. And, the ideas were generated very quickly, saving a ton of time to generate an early concept. However, it lacks the in-depth knowledge of your industry (e.g., sustainability and the circular economy) or company processes and expertise, the sudden burst of inspiration, or the polished “million-dollar idea” that has been in your head for years. And these tend to be the foundational elements of truly innovative design. However, these ideas only took 30 seconds to generate, and you can potentially get better results by evolving the prompt with additional details and your unique background information.

OK! Let’s move on to ideation, one of my favorite stages of design, and maybe the most flashy and cool way to showcase AI’s generative power.

• • Using Dall-e-2 (https://openai.com/product/dall-e-2)

And

• Midjourney (https://www.midjourney.com/home)

Prompt into Dall-e-2

“/Imagine A sustainable shoe recycled materials plastic, rubber, and other materials simple natural organic high tech minimal textured colors”

Prompt into Mid-Journey

“/imagine A sustainable shoe recycled materials plastic, rubber, and other materials simple natural organic high tech minimal textured colors”

Analysis

This is extremely impressive; it would take a designer many hours to create an image approaching this level of photorealism and detail. Generative AI art programs produce it in a matter of seconds and generate multiple variations of preferred directions very quickly. The trick to generative art is volume and chasing a specific vision. However, there is a pleasing chaotic factor to the images it produces that seem to come completely out of left field, some brilliant and some terrible.

Other than the prompt /Imagine, other commands include /Blend where you combine your own images. You have the ability to pull an image seed, keeping parts of the generated images’ DNA as you create images with new prompts.

Generative art is addictive; I have spent hours plugging away at the command line producing very cool and sometimes VERY weird results. The previous test inputs were generated quite quickly and still produced quite interesting images.

The primary downside I have found is that while you can create variations, you cannot take a sketch and refine it in the same useful way that a product designer can. It isn’t possible to assign where to put your logo or how to change a certain detail. It also doesn’t create working files that can be directly edited post AI, and the image cropping isn’t always logical or standard.

As we become more aware to AI art and design, we’re seeing quite amazing process examples, but it’s important to note that these users are taking the AI output and then cleaning them up through post-process. Things now are changing so quickly. In fact, a plugin for Photoshop was released while I was writing this paper, but it has not addressed this issue (yet).

The key problem is that of design intent. Product, brand, and graphic designers generally have a creative vision we are working towards. If you get stuck in the iterative art rabbit hole you can quickly lose your vision and your way. Moreover, AI generative art does not communicate the key functional design benefits clearly, and that is the main point of product design sketching. Sometimes you will find artifacts from the source randomly in the image.

To me, Dall-e-2 is the most approachable AI platform currently available. MidJourney was the first to get a lot of hype in the summer of 2022, has a vibe / style, and can generate very cool images, but the discord chat UI is a bit weird. I also like Dreamstudio for some of the tools, but it does not seem to generate as high quality of content as intuitively as other platforms. With as many of these AI art generators, I expect there to be a massive amount of innovation, feature releases, and product updates in the coming months and years.

Here at BOLTGROUP, we do see great potential in using AI art for inspiration and to create ideas that we might not have thought of otherwise. However, there will likely be several editorial steps to down-select and turn outputs into content that is useful to the design process and truly original to the project and client needs.

Another powerful capability is to create a loose sketch or rendering and use AI to iterate variations. That is not explored in this article but is potentially a process many designers will use in the near future.

Now, let’s dip into logo creation for CarbonFootprint.

Used :

• Adobe Creative Cloud Express Logo Maker
• Canva (www.canva.com)

Adobe Creative
Prompt into Adobe Creative

“CarbonFootprint sustainable footwear brand” and then select style from design guide.

Canva

“CarbonFootprint sustainable footwear brand” and then select style from dropdown menu.

Analysis

The logo creation tools seem a bit less open-ended than the generative art programs so far and tend to generate clean yet basic designs. Some new platforms have popped up since I started writing this post, so keep your eyes open. Logos are extremely hard to do well. Great logos are a brand’s fingerprint and often have personal meaning.

I would admit though, there are some decent hints at good ideas here. A graphic designer could use these as high-level concepts to build a deeper connection from the logo to the brand DNA.

Input to Chat GPT

Create social media post for “CarbonFootprint”—A sustainable shoe brand that aims to offset its carbon footprint by using renewable energy sources.

Output from Chat GPT
Introducing CarbonFootprint—The sustainable shoe brand that’s committed to reducing its carbon footprint! With renewable energy sources and sustainable materials, CarbonFootprint is leading the charge towards eco-friendly footwear. Each pair of shoes comes with a carbon offset certificate, showing how much carbon was offset during production. Join us in our mission towards sustainable fashion and a greener future. #CarbonFootprint #SustainableFashion #EcoFriendly #CarbonOffsetting #RenewableEnergy #Sustainability

Analysis
The copywriting tools are quite interesting and absurdly powerful. At this point, I would not trust it to create any final copy though the updates seem to have elevated the level of writing even more. They do write in proper grammar and spelling and can generate content in multiple languages. As mentioned earlier they can make up content or lie so it’s a good idea to verify any information you did not provide the AI. You can quickly rewrite, summarize, and generate a wide range of content almost instantly.

I think these tools are most useful when employed as a writing prompt, especially when dealing with writer’s block or when looking for a good place to start. For what it is worth, the first thing I did was try to get Chat GPT to write this article for me, but it was generic, vague, overpromised, and did not give specific examples of the benefits and downsides of AI.

For further context on the benefits, I have dyslexia and it is only using the AI writing assistant tool Grammarly that writing has become less painful in the last few months.

At BOLTGROUP we are extremely excited about the potential of AI and plan on continuing to integrate it into our workflow, but we will never use it at the expense of quality. How much creative content will ultimately be AI-generated vs human remains to be seen, but for now human designers will continue to be part of the process, bringing their skills to create innovative and empathetic human-centered designs that are sensitive to the needs of all stakeholders.

AI means there will be more design for more things than ever before, making it harder to stand out in an ocean of AI-generated content. Even more importantly, if everyone is creating massive amounts of somewhat random AI-generated creative content, what is the overall vision, the brand language that holds these disparate elements of brand, graphic, UI/UX, experience, and product design together? How does it relate to the people, the team members, and the customers?

So, do you still need a creative agency?

The answer is…Yes, now more than ever. But please explore these AI tools and get your imagination going.