Embedded Files
Roles: UX Content design · Conversation design · Synthetic data tailoring
Team: UX Content/Conversation Designer · Product Designer · UX Researcher · Product Manager · Engineering
I worked within the Google Assistant wearables team in the areas of content, conversation design, and product design for multimodality across devices and wearables.
Some of my responsibilities included assisting with fine tuning Gemini responses, creating synthetic data for training, and designing content to deliver customer experiences that are helpful, timely, and relevant.
Due to the confidential nature of projects, a lot of my work can't be discussed in detail.
Projects
Gemini Conversational AI response framework
Roles: Conversation design · Fine tuning · Synthetic data creation
Gemini, while being useful, was extremely verbose, given to overexplaining when asked even the simplest of questions. In addition to this, it also added information that wasn’t relevant to the information requested.
In many cases, users need short, simple answers to their queries due to device and/or environmental considerations when Gemini’s response can’t be the primary focus of their attention. In such cases, Gemini’s well-meaning verbosity could actually be a hazard, especially when used during activities such as driving or running. Additonally, overly verbose responses in audio-only scenarios could become annoying, leading to customer dissatisfaction.
Presented below is evidence of Gemini's verbosity. Note the overabundance of information.
Efficiently concise – Only the facts, please
The challenge here was to create a framework that would train Gemini to tailor its responses to be helpful, concise, and efficient while being informative across different use cases. It would need to be able to respond to the user’s prompt to the best of its ability without going off the rails with its response length.
I collaborated with a colleague to create a framework based on actual Gemini responses, along with tailored synthetic data. The ideal outputs lengths and the information they would contain covered the most common use cases, as well as important edge cases.
The actual framework and associated deliverables are covered by an NDA, but this visual demonstrates some of the kind of thinking that informed the creation of the framework.
Assistant features deprecation
Roles: Content design · Cross-functional collaboration
Google decided to discontinue some Assistant features. Among them was App Launcher in Android Auto that was related to driving mode for messages, calls, and controlling media. In order to communicate this change to users, I was tasked with creating original messaging to inform users about the upcoming deprecation, as well as post-deprecation messaging about alternatives they could use.
The key part of this project was that the messaging would show when App Launcher was being used while driving. This meant that the approach chosen would have to be such that the user was not put in danger, and should be able to quickly get the message without needing to pay too much attention.
This required the messaging to be extremely simple, short, and to the point. Verbosity had no place here. We communicated what the user needed to know and left it that, because saying more would possibly endanger our users.
So far, no crashes have been reported due to this messaging (whew!).
Google Sites
Report abuse