Answering Employee Questions with Natural Language Processing in IRWorks
Department of the Treasury · Internal Revenue Service (IRS)
Overview
- Development stage
- DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.
- High-impact designation
- Not high-impactSource: c) Not high-impact
- Impact justification
- The output is not presumed to be high-impact and is not used as the principal basis for significant decisions/actions
- Start date
- 2023-04-30 (day precision)Source: 4/30/2023
- Withheld from public reporting
- NoSource: a) No
Mission
- Topic area
- Information Technology
- Operational functions
- Search and knowledge retrieval Derived by keyword rules; see methodology
Problem the AI is intended to solve
The purpose of the Artificial Intelligence (AI) element of the IRWorks platform is to utilize natural language to respond to common user questions about IRWorks knowledge base (KB) articles and service catalogue items in the IRWorks Employee Center. This AI element is interpretive. At a high level, this is accomplished by using language models to train the agent to respond to user questions. The decision/judgement that this AI element uses is to search existing Knowledge Base (KBs)/service catalogue items, in conjunction with the user's defined access within IRWorks, to automate Tier 1 support by outputting (in natural language) answers and/or associated links that align to the customer's (natural language) query. Yes, humans are involved in reviewing the output - the AI element does not produce actionable items but produces information or links that are provided to the querying user for them to review and/or take manual actions. The population impacted by this decision is all IRWorks users, about 90K+ IRS employees.
Expected benefits
This functionality allows better understanding of user questions to direct them to better information allowing for quicker resolution of IT and HR queries and issues.
System outputs
Improve efficiency and optimize outputs for end users requesting services/information from within ticket / workflow management platform.
Technology
- AI classification
- Natural language processingSource: Natural Language Processing: AI that processes, interprets, and shares information in human language.
- System name(s)
- Not reported
- Custom-developed code
- No
- Public source code
- Not reported
Sourcing
- How it was built
- Purchased from vendorSource: a) Purchased from a vendor
- Vendor (as reported)
- Not reported
- Vendors (standardized)
- None on the standard list
A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.
Data and privacy
- Involves PII
- No
- Privacy Impact Assessment
- Not reported
- Authorization to Operate
- Yes
- Demographic features
- k) None of the above
- Training and evaluation data
- Knowledge Base Data for Information Technology Service Management (ITSM) including how to guides. IRS specific 'utterance' phrases that determine user intent. Vocabulary synonyms for words or phrases unique to IRS.
- Federal Data Catalog
- Not reported
Governance
0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.
Potential impacts and how they were identified
Not reported