In today's rapidly evolving digital landscape, creating an exceptional user experience is more crucial than ever. As we strive to design products that are not only functional but also delightful, integrating Artificial Intelligence (AI) has become a game-changer for UX designers. In this blog post, we’ll delve into how AI is transforming the field of UX design, enhancing creativity, efficiency, and personalisation to take user experiences to new heights.

An illustration that captures the practical application of AI in modern software development, with a focus on digital product consultancy. The image is in a futuristic style.

It’s likely that you didn’t notice anything strange about the paragraph you just read, even though it was written in mere seconds by the AI-powered chatbot, ChatGPT

Since it was released in November 2022, the ChatGPT website has clocked more than 10 billion all-time visits. People are using the tool to plan holidays, find jobs and write music. An offshoot feature of ChatGPT called DALL-E 3 can create impressive AI-generated artwork with just a simple text prompt. All this is pretty mind-boggling, to put it mildly. Users can access the basic version of ChatGPT for free, while the paid version costs US$22 per month. The release has spurred a rush of competitors such as Google Bard, Microsoft Bing Chat and Meta Llama. The AI news cycle continues to churn, as the CEO of the company that owns ChatGPT was recently fired and then reinstated within a week.  

How does ChatGPT work?

Online learning platform Growth Tribe describes ChatGPT as:

“A large language model (LLM) that uses natural language processing and machine learning algorithms to supercharge its Artificial Intelligence giving you pretty much accurate answers.”

Alternatively, we can lean on ChatGPT to describe itself. If we ask the chatbot to ‘explain how ChatGPT works in simple terms’ the following answer is generated:

Sure, I'll break it down in simple terms:

  1. Input: You start by typing something, like asking a question or stating something.

  2. Understanding: ChatGPT reads and tries to understand what you said. It doesn't have feelings or consciousness, but it's really good at recognising patterns in language.

  3. Context: It looks at the words you typed and the context of the conversation so far. It doesn't have memory like humans do, but it "remembers" the recent parts of the conversation.

  4. Generating a Response: Based on what it understands and the patterns it has learned from lots of text it was trained on, it generates a response.

  5. Output: It gives you a reply. The more information you provide and the clearer your input, the better the response is likely to be.

It's like having a very advanced and knowledgeable pen pal who can chat with you about almost anything, but it doesn't have personal experiences or emotions. It's just really good at talking in a way that seems quite human-like!

As you can see, ChatGPT does a pretty good job of explaining the process in simple and friendly terms. The answer it provided feels believably ‘human’, almost as if it were an explanation that a colleague had written for you. The difference is that instead of hassling a human, ChatGPT is on standby to answer your prompt in seconds. 

What are the concerns about ChatGPT?

Sometimes, ChatGPT appears to go rogue and starts ‘hallucinating’. There have been instances of the chatbot spouting transphobic hate speech or citing fabricated academic journals. More broadly, there are ethical concerns about the implications of AI being used without careful regulation, specifically when it comes to online data security and deepfake imagery creation. 

In Australia, National eSafety Commissioner Julie Inman Grant has written that “shaping our post-AI world will be entirely dependent on bold actions we are prepared to take today”. The agency has rolled out the Safety by Design toolkit, which provides globally accessible assessment tools to ensure that user safety is baked into digital products from the start.

At an international level, workers have been flooded with headlines warning that AI could absorb their roles. Investment bank Goldman Sachs reports that the equivalent of 300 million full-time jobs could be replaced by AI, particularly in administrative and legal sectors. 

How is AI being managed so far?

AI is such a hot-button issue that it has prompted responses from both the public and private sectors. Adobe has introduced Content Credentials to encourage creators to tag images that have been manipulated by AI. Google has gone a step further, including a visually invisible digital watermark in the metadata of AI-generated content. At the recent AI Safety Summit, 28 countries signed the Bletchley Declaration which affirmed that “AI should be designed, developed, deployed, and used, in a manner that is safe, in such a way as to be human-centric, trustworthy and responsible”. U.S. President Joe Biden also signed an AI Executive Order, the first of its kind. 

What does this mean for human-centred design?

When deployed judiciously, AI tools can function as significant time-savers throughout the human-centred design process. Human-centred designers can leverage AI to transcribe interviews, summarise insights, analyse usage data and generate imagery to facilitate storytelling.

It’s important to note that these tools work best when integrated as a layered approach:

  1. The designer first uses AI to surface initial themes or ideas as a starting point.

  2. The designer then applies their understanding of context, nuance and stakeholder requirements. 

While sifting through large datasets of research findings might be better suited to an LLM tool, other aspects of synthesis require a more nuanced or holistic understanding of a context or domain. Similarly, lateral thinking or making connections between seemingly unrelated ideas is sometimes also only possible using the experience a practitioner has across multiple previous projects.

Nomat is constantly looking for innovative ways to enhance our work processes and provide the most value to our clients. Currently, internal trials are underway to gauge the efficacy of AI tools when analysing and synthesising large datasets from research. These experiments are done using trusted tools with multi-factor authentication. Additionally, Nomat is in the process of creating an AI policy to be released later in 2024.

As the usability pioneer Jakob Nielsen says in his recent blog, start now but start small. Gradually introducing AI tools in the human-centred design process will allow for increased productivity, and an enhanced user experience. 

About the author

About the author

Eliza Crisp

Designer and Researcher

Eliza has a passion for investigative research and Human Centred Design with more than five years experience. She has a keen interest in applying Service Design to Health Care with a view of achieving improved outcomes for patients and staff. Eliza has strong communication skills, particularly when writing for the web. She is empathetic, collaborative and endlessly curious, and brings her enthusiasm for good design to every project.

Latest posts

Interested to know more? Let’s Talk.

Interested to know more? Let’s Talk.

Interested to know more?
Let’s Talk.