When asking them how UX experts are preparing for generative AI and how it’s already being used, Ameet rightly pointed out that AI is a bit of a buzzword right now and we should be focussing on gaining a better understanding of AI and its use cases.
At Transform, we're constantly investigating AI tools from an interaction, content and research perspective and have found there’s areas where it has impact. It can, as an example, support deeper research, provide variations of content or be used for storyboarding. Research isn’t just about looking at new tools, but also exploring the capabilities of the ones we already use. Atlassian Intelligence within Confluence is a perfect example because it summarises larger research documents before identifying the main points which can then be translated into potential design challenges later in the design cycle.
However, within interaction itself, AI remains more of an aid, an assistant which fits into our view that it’s not going to substitute what we do. It’s about taking mundane or data-heavy tasks and handing them to AI.
Gwyn added that a lot of the conversations around UX and generative AI touches on the idea that AI can replace the design process, by going off and building, for example, a full set of screens. AI, however, is missing the context and the feedback coming from key stakeholders or subject matter experts. If you think AI is going to design without any context, you'll end up spending a lot of time correcting all the mistakes it's made, taking longer than building it yourself. If, instead, you start using AI as an assistant for doing research and initial flows, it’s a fantastic tool that can speed up your work and keep you at pace – exactly what Emma Robertson, Transform’s CEO, touched on in her recent interview.
One step in Transform’s experimentation with AI has been looking at the concept of redesigning the driving licence renewal process on the GOV.UK website.
When tasked to understand the best UX approach involved in renewing a driving licence, Ameet told us he investigated how AI could help him in the brainstorming phase. AI text-prompt generated tools, for example, provide instructions to help you understand the best wording combinations for effective results.
When you don’t have access to the experts, inputting information into tools like ChatGPT could be very useful. You can also ask AI to design a set of screens and compare suggested designs. Alternatively, if you're looking to improve that process, it can provide the initial outline of the end-to-end user flow to understand potential gaps and solutions within that service.
Let’s not forget the human touch, though. It’s still our job to review the copy or ensure that it's over a certain reading age level, depending on the audience you’re developing it for, as well as find and add the details that AI might not necessarily have thought about.
Gwyn also pointed out that AI can do a lot of the thinking around internationalisation. In keeping with the driving licence example; if you wanted to collate information of existing driving licence-holders, we wouldn't necessarily know what an American driving licence looks like. That’s the sort of information AI can source. It might be that licenses are issued by the state and not the country, that the dates are formatted differently or that you need a zip code field instead of a post code. The user might have a different expectation or set of data and AI could prompt you to improve their experience.
A good example of how AI can reduce your workload came up when Gwyn talks around image creation. Up until now, if you wanted to create a storyboard with sketches, you had to take a picture of a space, go into Illustrator and hand trace it. Now, as you can imagine, that takes a long time.
With AI and only a few commands, instead, we can create a pencil sketch or visual in any style, having creative control and being able to quickly amend it if the storyboard changes. That control also makes it much easier for researchers to replay their findings and solutions by depicting it visually.
“It’s a learning process and sometimes like trying to coax squirrels with peanuts,” warned Gwyn. “You put in what you think are great text prompts, and the results make little sense. Other times you use text prompts which you think aren't clear enough and it fetches very effective results. It's down to which AI generator you're using and trying to find the right tool at the right time to give you what you want.
None of these tools are necessarily the right or the wrong ones to use, per se. They all have strengths; they all have weaknesses. It's about trying to work out what the AI vocabulary is to encourage the best results.”
Finally, we touched a little bit on accessibility, an incredibly important topic in every industry but particularly within the digital space. Accessible code means accessible services.
When talking about accessibility, it’s not only from a content perspective, but from a front-end coding perspective as well, so that a user can effectively use screen readers. Again, the human touch still plays into it, because you'd want to have someone with expertise to review and ensure that it's up to the right standard.
Here’s where tables turn, with the distinct possibility that interaction design will need to become more social and multimodal in character, as conversation, interaction with infographics and videos all become automatable. If we think about automation brought by AI and the need to always keep a human eye on it, you quickly realise that there’s a chance that you need more people rather than less.
At least for now, whilst AI is an array of tools and lacks the back story, UX designers' expertise are needed to make sure everything runs smoothly. In some ways, as UX may become even more complex, it might ironically require more workforce and the added value only consultant’s expertise can bring.
With AI quickly improving and moving forward though, the best approach is to face the fact that at times a certain tool might be ahead of the curve whilst falling behind at other times. Staying on top of changes will keep you in the loop on what other tools there are, and which are best to take advantage of in every situation.
If you’re interested in knowing more about our experiments with AI within UX, drop us a note at transformation@TransformUK.com and we’ll put you in contact with one of our experts. And if you’re in London on the 5th of March, why not popping at our stand at The Customer Experience Conference? Transform’s CEO, Emma Robertson, will share more insights on AI during a Q&A panel – you can read her thoughts ahead of the conference here.
