The rise and rise of AI in research and UX

As times rapidly change, it's easy to get carried away by the allure of new tools that seem as vast as the ocean. However, they may not be as deep as they appear.

These tools hold tremendous potential, promising innovation, and improved user experiences. But, it's important to approach them with a critical eye, considering their limitations and potential pitfalls, especially when it comes to understanding and brokering user data.

In environments where principles come before following any framework blindly, we are prompted to ask: What are the new methods of acquiring data in environments where principles come first? And, what ethics, morals, and biases do we need to learn to re-navigate such environments?

Democratizing research with Artificial intelligence

UX has long existed as a discipline to be done based on best practices, pattern-related studies, or empirical knowledge. Better UX though, is informed UX on a case-to-case basis. No matter the seniority, your experience, and general best practice knowledge might not be enough.

Working within the same industry but on different products doesn't necessarily make you an expert in starting a new product. Not only can the USP and niche change but the business goals could also shift. Most importantly, the people meant to use this product, might be different. That can take any professional down to the basics and start over. If you embrace it, there is a chance to do something new and innovate.

To be able to understand that UX does not happen to a product or within it but within the mind of the person using it immediately brings questions of understanding it. That person's mind and thoughts and feelings are as elusive as user research data can be. Qualitative or quantitative - there is always a person behind the number. With the beginning of mass AI usage, questions arise about how we can use it without oversimplifying the rich tapestry of human experiences.

AI does not influence our products directly. Instead, it affects us as humans, because that is where user experience is formed. UX is not created solely within or for a digital product but is ultimately perceived and processed within the human mind.

Some staples of using AI to improve UX already seem to be here to stay: personalized user experiences, intelligent search and content recommendations, accessibility and inclusive design, smart automation, and micro-interactions. Advancements in technology have positively impacted user research, leading to more informed UX. This progress is evident through predictive analytics, user behavior insights, automated user research and testing, and notably, natural language processing and conversational interfaces. The employment of custom databases is especially intriguing in the context of these interfaces.

There are many ways to experiment with custom knowledge bases and use them to build conversational chatbots. Some have done it manually. Others have used smaller tools like Cody or AskYourPDF for simpler solutions, and some are using powerhouses such as Google Dialogflow, IBM's Watson, or even Azure's QnA maker.

Analyzing data this way may have drawbacks. However, its forte is shaping up to be the way we present knowledge backing product-based decisions. Using a conversational interface allows for a democracy that was unachievable by limiting data analysis to a small team of people delivering results through presentation slides and charts.

We may be progressing towards empowering every team member with knowledge that not only facilitates decision-making with minimized risk but also help foster collaboration between multidisciplinary teams in an unprecedented way. This improvement in collaboration is due to a deeper understanding of the product.

Oversimplifying the complexity that is human behavior

AI tends to oversimplify complex human behaviors when automated user research processes are pursued. This leads to drawbacks in understanding overall user experiences.

While AI algorithms excel at identifying patterns and correlations in large datasets, they often struggle to capture the nuanced and multifaceted aspects of human behavior. This oversimplification can result in misleading or incomplete insights, hindering the ability to design truly impactful user experiences.

One of the limitations of AI-driven user research is its difficulty in capturing the subjective and context-dependent nature of user preferences.

Human experiences are influenced by a myriad of factors such as culture, personal history, and social context. These nuances often require deep qualitative exploration and human interpretation to fully understand.

AI, with its reliance on quantitative data, may overlook these subtle nuances, leading to design decisions that fail to resonate with users on a deeper level.

Furthermore, AI models may struggle to capture emotional responses and complex user sentiments accurately. While AI can analyze textual or visual content to some extent, it often falls short in recognizing sarcasm, irony, or subtle emotional cues that humans easily grasp. The inability to capture this can result in misinterpretations and ineffective design decisions.

To address the pitfalls of oversimplification, it is essential to complement AI-driven research with human expertise and qualitative research methods. Human researchers possess the ability to empathize, interpret, and delve into the underlying motivations and needs of users. By conducting in-depth interviews, ethnographic studies, and contextual inquiries, designers can gain a deeper understanding of user behaviors and preferences that AI algorithms may struggle to capture.

We can gain a more comprehensive understanding of user experiences by triangulating data from multiple sources. These sources include AI-driven insights and qualitative research. Integrating qualitative feedback and observations can uncover the "why" behind quantitative patterns, enabling designers to make more informed decisions.

What the future holds

As we have accepted our roles as validators and context-givers, it's interesting to see some early studies on generative writing tools' symbiosis with human performance.

A recent study conducted by MIT observed the productivity of professionals. They found that using ChatGPT to write business documents reduced task time. Additionally, the quality of the documents was rated as substantially improved.

That personifies how we observe most of what AI has to offer - faster work, and better results. But when looking at the study in more detail, it was found that using ChatGPT changed the way users spent their time.

It seems that when using ChatGPT, participants possibly spent less time brainstorming. Time spent generating rough drafts was more than cut in half since most of this workload was offloaded onto ChatGPT. And, interestingly, the time spent polishing the draft doubled (with no evidence that human editing is improving the ChatGPT output).

At the aggregate level, ChatGPT substantially compresses the productivity distribution, reducing skill inequality. The generative writing tool increases the output quality of low-ability workers while reducing their time spent. Meanwhile, it allows high-ability workers to maintain their quality standards while becoming significantly faster.

We might conclude that we'll probably keep spending less time ideating on our own, and more time providing context and validating AI-delivered content. All the while our experiences and expertise, no matter the level, might add up to similar quality outcomes.

Will our reduced efforts in the early ideation phases affect our sympathy to build meaningful products in the end?

Conclusion

While Artificial intelligence has the potential to revolutionize user research and enhance user experiences, it is essential to recognize and tackle its negative aspects and pitfalls. By remaining vigilant about biases, avoiding oversimplification, prioritizing privacy, and preserving human connection, teams can navigate the dark side of AI and ensure that user experiences are truly improved.

Striking a balance between AI-driven insights and human empathy is crucial to fostering ethical and inclusive user research practices that prioritize the well-being and satisfaction of all users.

At Martian & Machine, we consistently strive to enhance our data acquisition methods, driven primarily by the imperatives of efficiency, speed, and accuracy. The integration of artificial intelligence tools within our frameworks has proven instrumental in refining our approach. Our upgraded project toolbox will improve our approach with advanced AI tools. Keep an eye on this space, as we prepare to roll out these enhancements that hold the promise of propelling any project operations into an entirely new realm of possibilities.