On Human and AI Design

As a therapist who enjoys doing basic design, I am sharing my brief thoughts on therapists’ increased use of AI to generate websites, flyers, emails, and other marketing materials.

Note: I am 100% in support of folks using website builders to make their own websites. This notice applies not to people who manage these roles themselves, but those instead tempted to have generative AI fully fill the role of website designer.

As AI continues to evolve, the points made here may cease to be accurate or relevant, with the exception of the ethical considerations.

AI Website Design

If you’ve been eyeing recent advances in technology and are considering generating your website with AI, there are reasons to pause. I always encourage awareness of the risks and the benefits in our options. 

A brief sample of issues I have personally seen with these sites is as follows:

It is apparent that they are AI-generated to those familiar with AI, meaning there are noticeable differences from a site that a human designer would make.

I would actually argue that this is the most significant point to consider, which I will touch on briefly at the end of this list.

Additionally, AI regulation increasingly requires AI-generated content to be identifiable, and one way this is done is by coding patterns for the AI to follow. Even outside of these regulations, AI tends to follow model-specific patterns that differ from, say, following best practice design guidelines.

In contrast, similar themes in a human designer’s portfolio helps you get an idea of what your site can look like, and if you do not like their style, you can browse other designers’ portfolios. 

Without human editing after content generation, the site can be generic. In the long run, the site not being specific to you will be detrimental as more and more sites begin to look and say the same.

While an issue of sameness was already apparent with blog posts written for therapists by marketing services (which, by the way, marketing copywriters often use AI, and prior to that, relied on templates), think of it as if all of those therapists had the same copywriter making slight revisions on the same article.

You can tell a human designer in depth context about your practice, and you can provide references that the designer can actively conceptualize to create a layout that complements your business plan.

AI tools, especially those not integrated solely for website creation, generate broken sites akin to the infamous human hand issue.

Some of the issues are immediately noticeable, and others less so, ranging from floating webpages, excessive blank space, and odd content order to beautifully arranged blog post excerpts that link nowhere because the posts do not exist.

These errors are very AI-specific and will not be found among typical human errors.

Although the site might look visually appealing, the accessibility is very low. Common issues I have noticed are:

  • Transparent backgrounds and light text that, in theory, are great, but on execution don’t account for the backgrounds that show through the transparency.
  • Small boxes that intend to make information digestible, but there’s an uncurated and excessive amount of these boxes, negating the purpose of the boxes.
  • A lack of organization of the information or use of proper hierarchy, making it difficult to find relevant information.

There are tools you can use to assess the accessibility of your site; however, accessibility is also a native element of graphic design. AI attempts to execute some of these guidelines, but it seems to struggle with determining which guidelines are most relevant and how best to combine them.

The sites have errors that are less likely to be caught because they were not created and reviewed by a person.

Typical errors on a human-designed site would include typos, formatting errors, and forgotten information. In addition to the other issues addressed here, AI can hallucinate and misinterpret prompts, make up information that it was not provided, and change information that it was.

Errors like hallucinated contact information or copy that implies things you did not mean are less likely to be noticed. If you use generative AI, you should always check links to make sure they work and lead where expected.

Impact

Without numbers to show it, I would suspect the above has an impact on image. A nicely formatted website improves your image and immediately says “professional,” but a good website also includes functionality and compliance, and communicates your personality or practice.

Pushes for rapid technological advancements and fights for regulation are directly impacting the way we network and exchange referrals. Therapists create impersonal AI-generated networking emails or ISO posts with dead links. Clients land on a pretty website with no way to connect because there is no contact form and the number on the site is fake. Unfortunately, these limitations can produce judgments or reflect aspects of the therapist that then lose leads.

Ethics

On top of this, there is a cultural competence component—generative AI is conceptualized differently across populations and subgroups. Being aware of the potential impact of AI use in your practice is a core aspect of being an ethical therapist. While you may not serve populations of clients who are wary of AI or who do not have access to information for proper informed consent, this is also an area that could be a blind spot.

We additionally have a responsibility to maintain our own awareness of the technology impacting our field—that does not simply mean embracing it as inevitable and therefore ignoring evidence of real issues. It means knowing the risks and benefits, engaging in informed use, and limiting or barring that use where ethics dictate.

AI is not a mandated tool for many of us, and thus the struggle is not between regulations and ethics but instead between ethics and ourselves. While I believe it is possible to be well-informed, write strong prompts, and use a refined tool with a lower error rate, as more therapist sites, flyers, and emails are AI-generated, crafting a generative website that does not run into the issues above will require more work and research. Often, the issue is that AI is used as an easy generation tool, so people do not review and edit the content. AI may also be expected to bridge the gaps of what the user does not know, meaning the user does not have the skills needed to identify and make corrections.

Larger Issues

The risks and benefits do not only touch our field, and impacts outside of our field are also relevant because they concern our clients. Changes outside of the field can have indirect and eventually direct impacts.

For example, there is valid commentary on larger issues regarding how AI is affecting creatives. The decision to be an artist, and the work artists produce, have both been consistently devalued societally. Artists have struggled with theft, negative commentary about pricing, and attempts to replace their labor when it cannot be obtained for free.

At a point where many people are using Canva to make quick flyers or AI to make cluttered lookalike flyers, adding a one-time designer fee to your budget for tailored marketing products ironically is now the thing that will make you stand out.

I won’t ping pong into another topic, but there are many people unfortunately posing as designers now who will charge you, and then input a prompt to create something via AI, and mislead you that they created it themselves. Luckily, most artists are happy to point people to legitimate human designers because this issue is well-known in the artist community.

Choice

AI tools are expanding to also include website maintenance, with sites like WordPress including them in their website builder, so it is entirely up to your discretion to choose that path. Ultimately, it is important to continue to be curious, challenge ourselves, and stay on top of auditing AI output.

Scroll to Top

Interested in working together?