One of the key benefits to managing and utilizing data more efficiently is that it enables the use of advanced capabilities such as artificial intelligence (AI) and machine learning (ML). The Department of Labor has recently begun exploration of how these advanced technologies can be used to benefit the agency and help deliver on our mission. In an effort to create transparency in the adoption of these tools, this page serves to highlight the various uses of AI across the department.
Active AI Use Cases
| Use Case Name | What is the Intended Purpose and Expected Benefits of the AI System? | Stage of Development |
|---|---|---|
| Use Case Name Form Recognition Model for Benefits Forms | What is the Intended Purpose and Expected Benefits of the AI System? Custom machine learning model to extract data from complex forms to tag data entries to field headers. The input is a document or scanned image of the form, and the output is a JSON response with key/value pairs extracted by running the form against the custom trained form recognition model. | Stage of Development Operation and Maintenance |
| Use Case Name Language Translation | What is the Intended Purpose and Expected Benefits of the AI System? To automatically translate unofficial documents into various languages quickly and easily, without the time and expense needed for human translation. | Stage of Development Initiated |
| Use Case Name Audio Transcription | What is the Intended Purpose and Expected Benefits of the AI System? Transcription of speech to text for records keeping using natural language processing models. | Stage of Development Operation and Maintenance |
| Use Case Name Text to Speech Conversion | What is the Intended Purpose and Expected Benefits of the AI System? Text to speech (Neural) for more realistic human sounding applications using natural language processing models. | Stage of Development Operation and Maintenance |
| Use Case Name Claims Document Processing | What is the Intended Purpose and Expected Benefits of the AI System? To identify if physician's note contains causal language by training custom natural language processing models. | Stage of Development Retired |
| Use Case Name Website Chatbot Assistant | What is the Intended Purpose and Expected Benefits of the AI System? The chatbot helps the end user with basic information about the program, information on who to contact, or seeking petition case status. | Stage of Development Operation and Maintenance |
| Use Case Name Data Ingestion of Payroll Forms | What is the Intended Purpose and Expected Benefits of the AI System? Custom machine learning model to extract data from complex forms to tag data entries to field headers. The input is a document or scanned image of the form, and the output is a JSON response with key/value pairs. | Stage of Development Retired |
| Use Case Name HoloLens | What is the Intended Purpose and Expected Benefits of the AI System? AI used to train Inspectors to visually inspect unsafe areas from a safe location. | Stage of Development Operation and Maintenance |
| Use Case Name DOL Intranet Website Chatbot Assistant | What is the Intended Purpose and Expected Benefits of the AI System? Conversational AI Assistant & DOL intranet websites to help answer common procurement questions, as well as specific contract questions. | Stage of Development Operation and Maintenance |
| Use Case Name Official Document Validation | What is the Intended Purpose and Expected Benefits of the AI System? AI detection of mismatched addresses and garbled text in official letters sent to benefits recipients. | Stage of Development Retired |
| Use Case Name Electronic Records Management | What is the Intended Purpose and Expected Benefits of the AI System? Meeting NARA metadata standards for (permanent) federal documents by using AI to identify data within the document, and also using NLP to classify and summarize documents. | Stage of Development Initiated |
| Use Case Name Call Recording Analysis | What is the Intended Purpose and Expected Benefits of the AI System? Automatic analysis of recorded calls made to Benefits Advisors in the DOL Interactive Voice Response (IVR) center. AI is not used for analysis; AI is used only for transcription. | Stage of Development Operation and Maintenance |
| Use Case Name Automatic Document Processing | What is the Intended Purpose and Expected Benefits of the AI System? Automatic processing of continuation of benefits form to extract pre-defined selection boxes. AI tool will extract data from the forms. | Stage of Development Operation and Maintenance |
| Use Case Name Automatic Data Processing Workflow with Form Recognizer | What is the Intended Purpose and Expected Benefits of the AI System? Automatic processing of current complex workflow to extract required data. | Stage of Development Acquisition and/or Development |
| Use Case Name Case Recording Summarization | What is the Intended Purpose and Expected Benefits of the AI System? Using an open-source large language model to summarize publicly available case recording documents which are void of personal identifiable information (PII) or any other sensitive information. This is not hosted in the DOL technical environment and is reviewed by human note takers. | Stage of Development Retired |
| Use Case Name OEWS Occupation Autocoder | What is the Intended Purpose and Expected Benefits of the AI System? The input is state submitted response files that include occupation title and sometimes job description of the surveyed units. The autocoder reads the job title and assigns up to two 6-digit Standard Occupational Classification (SOC) codes along with their probabilities as recommendations for human coders. Codes above a certain threshold are appended to the submitted response file and sent back to states to assist them with their SOC code assignment. | Stage of Development Operation and Maintenance |
| Use Case Name Scanner Data Product Classification | What is the Intended Purpose and Expected Benefits of the AI System? Classifies bulk data received from corporations into Entry Level Item (ELI) codes in the Consumer Price Index (CPI) | Stage of Development Operation and Maintenance |
| Use Case Name Consumer Expenditure Diary Autocoder | What is the Intended Purpose and Expected Benefits of the AI System? Assigns expense classification categories to reported expenses from Consumer Expenditure Diary Survey respondents | Stage of Development Operation and Maintenance |
| Use Case Name Generative AI Assistant | What is the Intended Purpose and Expected Benefits of the AI System? Private and secure in-house solution to evaluate business use cases that can solve problems using Generative AI models and semantic search. Example use cases include text summarization, text analysis, and document comparisons. | Stage of Development Operation and Maintenance |
| Use Case Name Occupation Code Suggestion for Job Duties Data | What is the Intended Purpose and Expected Benefits of the AI System? Suggest appropriate occupation code for job duties data using natural language processing and classification techniques. | Stage of Development Retired |
| Use Case Name Notice of Deficiency (NOD) Text Generation using Generative AI Model | What is the Intended Purpose and Expected Benefits of the AI System? Application of custom Generative AI model to create Notice of Deficiency (NOD) text based on the input data. | Stage of Development Initiated |
| Use Case Name PII Redaction | What is the Intended Purpose and Expected Benefits of the AI System? Using Amazon Web Services Personal Identifying Information scrubber for ITA text fields and Named Entity Recognition to remove additional names from the text fields. | Stage of Development Operation and Maintenance |
| Use Case Name Website Chatbot Assistant | What is the Intended Purpose and Expected Benefits of the AI System? The chatbot helps the end user with basic information about the Workforce Recruitment Program and information on who to contact. | Stage of Development Operation and Maintenance |
| Use Case Name Initial Determinations for Incoming Applications | What is the Intended Purpose and Expected Benefits of the AI System? AI tool to help perform initial analysis of the incoming 9141 and 9089 applications. Make initial determinations regarding whether issuance of a Request for Information is necessary. Create a standard RFI request and either (1) automatically issue the RFI before case assignment and hopefully obtain customer responses by the time the case is assigned OR (2) perform this action at case assignment and allow the assignment analyst to affirm and issue the RFI or not. | Stage of Development Initiated |
| Use Case Name AI Course Design Assistant | What is the Intended Purpose and Expected Benefits of the AI System? AI tool to recommend the structure of a course, titles for modules, descriptions, and images. AI-powered algorithms analyze course content and quickly generate test questions and prompts for authentic assessments. (Blackboard/COTS Product, unmodified) | Stage of Development Operation and Maintenance |
| Use Case Name Worker PLUS Microsimulation Program | What is the Intended Purpose and Expected Benefits of the AI System? The Worker PLUS model was developed as an evolutionary iteration of the Paid Family and Medical Leave. Simulator Model developed by Albelda and Clayton-Matthews (the ACM model). | Stage of Development Operation and Maintenance |
| Use Case Name Computer-Assisted Coding: SOII Autocoder | What is the Intended Purpose and Expected Benefits of the AI System? The Survey of Occupational Injuries and Illnesses (SOII) collects hundreds of thousands of narratives describing cases of work-related injury and illness annually. Using narratives and other relevant information, SOII Autocoder automatically assigns classifications for SOII elements, which include worker occupation, nature of the injury, part of body affected, event that resulted in the injury, source and secondary source (if it exists) that caused the injury. The use of SOII autocoder initially began in 2012 for review purposes, then gradually expanded to automatically assign these classification codes. In reference year (RY) 2022, 92% of all SOII elements were automatically coded, which were then subsequently validated by human staff. The SOII Autocoder is a transformer-based text classification model using millions of SOII cases as the labeled training data and a publicly available, third-party language model as a pre-trained base model. The Autocoder is trained annually in house using a GPU server that is owned and maintained by OCWC. Once trained, the Autocoder is deployed internally via REST API, autocoding batches of SOII and MSHA cases on a routine basis during the production cycle, all within the BLS network. | Stage of Development Operation and Maintenance |
| Use Case Name CFOI Record Matching | What is the Intended Purpose and Expected Benefits of the AI System? The Census of Fatal Occupational Injuries (CFOI) collects and publishes a complete count of work-related fatal injuries and descriptive data on their circumstances. The CFOI Record Matching programs matches records from various sources, including Occupational Safety and Health Administration (OSHA) Information System (OIS) files, news articles, and Quarterly Census of Employment and Wages (QCEW), to identify missing or inconsistent data in CFOI. The CFOI Record Matching programs use various aspects of AI in the areas of natural language processing and record matching. The CFOI-OIS matching uses random forest classifier model trained on previously matched CFOI-OIS data. The CFOI-QCEW matching uses TF-IDF vectors and cosine similarity to match establishment names in the CFOI-QCEW data. And the CFOI-CPDMS matching uses a publicly available, third-party question-answering language model to identify relevant CFOI elements from CPDMS data and then uses random forest classifier model to match CPDMS data to CFOI data. These processes are all conducted within the BLS network, and the outputs, in the form of spreadsheets, are provided once or twice a year to the staff in the CFOI program for review. These processes are all run within the BLS network. | Stage of Development Operation and Maintenance |
| Use Case Name OIICS Coding | What is the Intended Purpose and Expected Benefits of the AI System? Auto-code assigning Occupational Injury and Illness Classification (OIICS). | Stage of Development Initiated |
| Use Case Name AI Assisted Coding - Microsoft GitHub Copilot | What is the Intended Purpose and Expected Benefits of the AI System? Automatically generate software code using Microsoft Copilot, this will reduce the time for generating code. | Stage of Development Operation and Maintenance |
| Use Case Name CPS OTC Prediction | What is the Intended Purpose and Expected Benefits of the AI System? The BLS productivity office publishes measures of hours worked by major sector and industry. As part of the calculation for this measure, BLS uses Current Employment Statistics (CES) data on hours paid is used to estimate hours for payroll workers. Additional adjustments are made to this measure, such as removing paid time off (PTO), adding off-the-clock (OTC) hours, and adding in hours worked by self-employed and unpaid family workers. BLS uses the Current Population Survey (CPS) dataset to identify the amount of hours that are worked off-the-clock (OTC) by workers. There are some workers in the CPS dataset that do not report whether their time off was paid, or whether they get paid hourly or not. This data needs to be imputed to calculate the ratio of total hours worked to paid hours worked. A random forest model is used to predict the responses for workers that did not report this information by training it on characteristics (such as industry, occupation, education level, age, etc.) of respondents that have reported that information. The benefit of this AI use case is increased accuracy. | Stage of Development Operation and Maintenance |
| Use Case Name Sample Refinement: Frame API | What is the Intended Purpose and Expected Benefits of the AI System? BLS establishment-based surveys, such as the Survey of Occupational Injuries and Illnesses (SOII), the National Compensation Survey (NCS), and the Occupational Requirements Survey (ORS), use the Longitudinal Database (LDB) as the frame for their survey samples. For each of these programs, there is a significant time-lag between when the sample is drawn for a program and when data collection begins. A critical and time-consuming part of data collection is sample refinement: checking the latest Quarterly Census of Employment and Wages (QCEW) and LDB data to adjust for any changes to the sampled establishment during that time-lag. The QCEW/LDB Frame API allows fast retrieval of the latest QCEW/LDB data for a nationwide sample so that users can perform sample refinement more efficiently. The SOII program uses the Frame API to perform over 20 sample refinement comparison checks, which outputs a report for the data collectors. The Frame API uses TF-IDF vectors and cosine similarity to compare company names, mailing addresses, and unit descriptions between the survey sample and the latest QCEW/LDB data to determine what components may have changed. The Frame API also has the flexibility for the users to match records using other combinations of variables. These processes are all run within the BLS network. | Stage of Development Operation and Maintenance |
| Use Case Name Consumer Expenditure Interview Item Code Estimation | What is the Intended Purpose and Expected Benefits of the AI System? Recommends an expense classification categories for Consumer Expenditure Interview Survey expense responses | Stage of Development Operation and Maintenance |
| Use Case Name Consumer Expenditure Interview Survey Imputations | What is the Intended Purpose and Expected Benefits of the AI System? Imputes missing expenditure values in the Consumer Expenditure Interview Survey when respondents answer “don’t know” or refuse to provide an expense amount | Stage of Development Operation and Maintenance |
| Use Case Name Note Taking Bot | What is the Intended Purpose and Expected Benefits of the AI System? The bot will take the transcript and summarize the meeting notes. | Stage of Development Initiated |
Contact Us
U.S. Department of Labor
200 Constitution Ave NW
Suite N-1301
Washington, DC 20210
Chief AI Officer: Mangala Kuppa
Email: zzOCIO-AI@dol.gov