One of the tools that got my attention a while ago is Midjourney as a result of noticing astonishing images made by an artist I know.

I only started using it recently and I have been surprised by the potential it has and at the same time it has forced me to think about the pros and cons of AI.

It is described by its creators as by its creators as “text-to-image AI where your imagination is the only limit”. One of the things I was curious about is the possibility to use Midjourney to create specific images that could be use in online courses or as feature images in posts as I sometimes struggle to find the right image.

I got myself a basic account to do some tests and I have selected some of the Midjourney generated images below to illustrate this post and the featured image.

As an instructional designer, part of the job may involve creating or editing certain creative assets. Most of the Authoring eLearning tools would nowadays provide access to a library of images, yet sometimes you may need something very specific. Same applies for blogs, there are a good number of bank photos but you may not always find what you want or you may spend too much time scrolling through a content library.

Creating an image for an online course

Let’s say that I need an angry dinosaur having a discussion to illustrate a conflict management slide for a course…

Prompt: angry customer with a dinosaur face discussing with a customer assistant, v 5.1 –style raw

If you look at the woman’s character, do you notice some strange things? The pen seems to be floating, the elbow and the hand are a bit odd.

And…the dinosaur body is…well…interesting.

This was the first attempt at this prompt which shows some of the AI limitations at the moment. Usually, by iteration and clarifying the instructions within the prompt, the image can get closer to the idea I had in mind although there is a high chance that it may still require some manual editing. On the positive side, if there is no similar image via the image bank, this can be very helpful.

Created with Midjourney

Creating a flat illustration

Now, using a more elaborated prompt to create a flat illustration, the result was not bad although it took a few iterations to get something that I was satisfied with. Still, the right arm looks like a bit strange to me.

Prompt: Discover the world of scientific exploration with this flat illustration featuring a chemist wearing glasses, with a smile and happy eyes, diligently working in a research laboratory. Engaged in experiments, she skillfully handles a [object that is used]. This minimalist UI illustration, perfect for GUI applications, boasts a clean white background and a bright color scheme that highlights the focus and determination of the scientist. Immerse yourself in the pursuit of knowledge and celebrate the passion for discovery in this vibrant and engaging visual representation.–ar 1:1 –v 5 –s 100

Created with Midjourney

The fun side of AI

So far, I have found being able to communicate with AI fascinating. It can be exciting when you manage to create a prompt that returns what you expected or frustrating when you get just the opposite and even funny when you look at some of the results based on not good enough prompts or the AI not quite understanding what you meant in your prompt (as you may note looking at the image below!).

Prompt: funny t-rex dinosaur, working as it support with headphones sitting in a chair in front of a computer, cartoon style, happy face, with scottish kilt ,v 5.1 –style raw –

Created with Midjourney

Like many radical innovations, the fear is for a potentially good tool to end up being used in a harmful way. It feels like a tsunami and either you sink or you ride the wave. I believe that, a good starting point, is to have a better understanding of its advantages and limitations, so that we can work towards and ethical and safe use of AI.

Thoughts about AI and Midjourney

Some quick thoughts based on my initial experience (note that I am not an AI expert or a designer)

Pros

• You can create almost anything that you can conceive in your mind
• It enhances your creativity by exposing you to creative options you may have not considered
• It can speed up the creative process using AI as a starting point and then adding manual input to create the final artwork.
• It can be used as inspiration, as creative prompts.

Cons

• Achieving a perfect image out of a prompt is not always possible, it requires iteration and often some final human touch.
• Complexity of copyright ownership/infringement (in theory, there is no copyright in the “raw” outcome as there is no creative human input unless it is modified, there are risks based on regulations in different countries, changes in terms of service, existing images…)
• Current controversy with copyright (it is worth reading Midjourney Terms of Service  and to check or seek advice depending on the intended usage of the images, especially for commercial use)
• Devaluation of manually created artwork

Opportunities

• Learning how to communicate to AI and train it to help us
• A powerful additional brain that can be used for good

Challenges

• Data usage
• Data waste as a result of the crazy amount of both good and awful content being generated.
• Concerns about Midjourney stealing work .
• Ensuring that we are able to establish ethical boundaries so that we do not end up fighting against Skynet like in Terminator!

Lessons learnt

• Talking to AI is not so different than talking to humans (bear with me…). If you want AI or a human to understand you, you need to communicate the right way. Also, even if you use the right words, AI may not get what you ask for!
• There is a learning curve, it can be frustrating and also good fun.
• To achieve the best results,  i would say that, at least at the moment, human input is required to edit what AI presents.
• The key to create successful prompts is to use a positive language!  If I want a “clear background”, the trick is adding “clear background” in the prompt rather than requesting “no dark background”. In essence, anything we may not want in the image, has to be changed into a positive keyword for Midjourney to better understand what we want.
• It requires a good conversation regarding copyrights and commercial use of creative work produced in collaboration with AI tools.

Final thoughts

• Although this AI tool is pretty amazing from a creative perspective, I have found that it can be time consuming to design or adapt certain prompts (hence the rise of prompt designers). Therefore, I may have spent more time trying to achieve a decent image with the prompts than finding a good enough image through the image bank or image library. It is also likely that a prompt designer may be able to achieve better results in less time.
• With the right knowledge and experience, it allows creating fascinating art and creative assets that otherwise, unless being an artist or a creative, may be near impossible to achieve.
• It can be a complementary tool to create certain digital assets that may be difficult to find and may be used as a first draft before editing the image through human input.
• There are some Midjourney users creating fascinating pieces of work however I don’t feel it is a tool that anyone can use to create a great creative asset in five minutes. Even knowing about prompt design, it seems that it requires some good skills to achieve really good outcomes.

If you feel like giving it a try you will need a Discord and Midjourney account. There are several good tutorials online in case you may want to give it a go.

AI, Instructional Design and Midjourney / EdTech Unchained by is licensed under a