Insights
How LLMs are reshaping digital experiences
An exploration into how large language models can support and enhance traditional software experiences
By
Jaden Flores
—
11
Jul
2024
Large Language Models (LLMs) are deep learning algorithms that ingest huge amounts of data to understand, predict, and generate new content. These systems have the ability to understand and interpret human language; their real-world applications are expected (and have already begun) to go beyond traditional conversational interfaces. As user experience designers, we’re drawn to these new technologies and their potential to improve digital experiences.
Interacting with LLMs
Language models aren’t new, but recent advances in natural language processing (NLP) have unlocked new opportunities, making LLMs an exciting space for design. As they become more relevant to our own process and to our partners, we've identified a few ways you can interact with LLMs.
Direct conversations
In their most familiar form, LLMs are presented through a simple chat interface where the user interacts directly with the model. These interfaces are prompt- or query-based. With a blank canvas, users can create their own creative path for the interaction, limited only by their imagination.
Contextual analysts
LLMs can also act as contextual helpers and analysts, allowing users to interact with tables and charts, draw insights, and make sense of their data. These types of applications take foundational models like Generative Pre-trained Transformers (GPTs) and feed them additional context in the form of data sets, PDF documents, and more. The LLM’s natural language processing abilities and intuitive understanding of data enable users to explore their data, answer complex research questions, and accelerate everyday tasks.
Behind the scenes assistants
For a more efficient workflow, LLMs can help automate tasks behind the scenes and shed light on the nature, intent, and tonality of information. These models can generate contextually relevant text and concise summaries, detect sentiment expressed in data, and categorize names, places, and companies.
The information synthesized by the model helps to inform the UI. So, instead of interacting directly with the LLM, teams can use apps like Grain to capture information discussed during a meeting and receive a well packaged summary of key points, transcripts, and more.
Collaborative agents
As AI continues to advance, we’re also seeing the rise of Agentic UX, where LLMs act as teammates or partners with a specific personality and purpose. With agents, the user’s workflow becomes less of a direct question and answer interaction, and more iterative and conversational. Agents can help users complete more complex tasks by serving as subject matter experts who proactively make suggestions.
In a recent project, we collaborated with Fixify to change the face of IT by supporting analysts with AI. We expanded on the concept of digital sidekicks, or conversational help desk agents, to create a collaborative interface where agents exist in context to streamline workflows.
The transformative power of LLMs
LLMs are drastically changing the way we access and analyze information. Whether through conversational UI, agents, behind the scenes assistants, or contextual analysts, users can describe a problem or task and leave the rest up to the capabilities of the system. This ability to break down complex queries suggests a new era for user experiences, where the system can quickly understand what the user needs and provide personalized answers and recommendations in real-time.
The transformative power of LLMs lies in this personalization. The collaboration between the user and the system creates a new kind of experience where the interface is dynamic, helping to gain insights, offer time-saving solutions, and improve overall user engagement.
What’s next?
In an industry where nothing stands still, it’s hard to know what lies ahead. Though, at the rate AI is advancing, it's highly likely that LLMs will become more seamless and integrated in the future. We're already seeing this with Google's recent search engine update, AI Overviews, which uses AI to provide more accurate and relevant results. Rather than being marketed as a separate feature or add-on, AI will start to become more native and integrated into products.
While it's exciting to experience this new paradigm of design, it's important to stay true to our user-centered roots. User experience starts with the user and is based on the idea that technology should solve real problems and user needs. Consider why the user is coming to your product, and whether it’s usable and responsible. As John Maeda says, “design humanizes technology.”
Further reading
Jaden Flores
With a curious approach to design, Jaden has crafted thoughtful words and experiences for companies like Bose, Experian, and Visa. As the resident grandma of the studio, Jaden enjoys taking walks, browsing local bookstores, and crocheting at home.
