Satisfying UI and Product Requirements and Improving User Experience, Trust, and Adoption

Table of Links

Abstract and 1 Introduction

2 Survey with Industry Professionals

3 RQ1: Real-World use cases that necessitate output constraints

4 RQ2: Benefits of Applying Constraints to LLM Outputs and 4.1 Increasing Prompt-based development Efficiency

4.2 Integrating with Downstream Processes and Workflows

4.3 Satisfying UI and Product Requirements and 4.4 Improving User Experience, Trust, and Adoption

5 How to Articulate output constraints to LLMS and 5.1 The case for GUI: A Quick, Reliable, and Flexible Way of Prototyping Constraints

5.2 The Case for NL: More Intuitive and Expressive for Complex Constraints

6 The Constraint maker Tool and 6.1 Iterative Design and User Feedback

7 Conclusion and References

A. The Survey Instrument

4.3 Satisfying UI and Product Requirements

4.3 Satisfying UI and Product Requirements Respondents stressed that it is essential to constrain LLM output to meet UI and product specifications, particularly when such output will be presented to end users, directly or indirectly. A common case is to incorporate LLM-generated content into UI elements that “cannot exceed certain bounds”, necessitating stringent length constraints. Content that doesn’t “fit within the UI” usually gets “thrown away” all together, a concern likely to be more pronounced on mobile devices with limited screen real estate [6, 20]. Maintaining the consistency of output length and format was also considered important, as “too much variability in the generated text can be overwhelming to the user and clutter the UI.” Moreover, being able to constrain length can help LLMs comply with specific platform character restrictions, like tweets capped at 280 characters or YouTube Shorts titles limited to 100 characters.

4.4 Improving User Experience, Trust, and Adoption

Finally, respondents suggested that developing LLM-powered user experiences requires constraint mechanisms to mitigate hallucinations, foster user trust, and ultimately drive “user adoption.” One prominent aspect is to reduce safety and privacy concerns, for instance, by preventing LLMs from “repeat[ing] existing or hallucinat[ing] PII (personally identifiable information).” In addition, respondents expressed a desire to ensure user trust and confidence of LLM-powered tools and systems, arguing that, for example, “hallucinations in dates are easy to identify” and, in general, “users won’t invest more time into tools that aren’t accurate.”

\

:::info This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license.

:::

:::info Authors:

(1) Michael Xieyang Liu, Google Research, Pittsburgh, PA, USA ([email protected]);

(2) Frederick Liu, Google Research, Seattle, Washington, USA ([email protected]);

(3) Alexander J. Fiannaca, Google Research, Seattle, Washington, USA ([email protected]);

(4) Terry Koo, Google, Indiana, USA ([email protected]);

(5) Lucas Dixon, Google Research, Paris, France ([email protected]);

(6) Michael Terry, Google Research, Cambridge, Massachusetts, USA ([email protected]);

(7) Carrie J. Cai, Google Research, Mountain View, California, USA ([email protected]).

:::

\