Learning Objectives

After completing this lecture, you will be able to:

• Understand the key concepts of ai in government today in a government context

• Connect ai in government today to your agency's AI initiatives

• Identify next steps for applying these concepts in your role

Key Topics Covered

• Current use cases across federal agencies (VA, IRS, USPS, DoD, EPA, SSA)
• State and local examples
• Interactive map of government AI deployments

Why This Matters for Government

Government agencies face unique challenges when it comes to AI adoption. This lecture addresses these challenges head-on by providing all government employees with the knowledge and frameworks needed to navigate AI in the public sector responsibly and effectively.

As part of the L1 (AI Aware) curriculum, this lecture builds on the foundational principle that every AI system in government ultimately serves citizens. Whether you are working with AI tools daily or setting strategy for your agency, understanding ai in government today is essential for responsible, effective government AI adoption.

======================================================================

TRANSCRIPT: AI in Government Today

======================================================================

What you will learn: Real AI deployments across federal, state, and local government. What's working. What failed and why. Where your agency fits in the landscape.

Theory meets practice. You now understand how AI works and what it's good for. But what does this actually look like in real government? This lecture takes you through the landscape: what agencies are doing with AI today, what's working, what went wrong, and what you might encounter in your own workplace.

You'll see federal agencies from Veterans Affairs to the IRS, EPA to DoD. You'll see state innovations and local experiments. Some are brilliant. Some are cautionary tales. All are instructive.

Purpose

Understanding the current state grounds abstract concepts in reality. When your agency proposes an AI initiative, you'll recognize patterns from successful deployments elsewhere and avoid mistakes others have already made.

Why This Matters for Government

Government AI is not hypothetical anymore. It's deployed. It's affecting citizens. Understanding what's actually happening is essential for your credibility and your agency's success. You need to know: What are peers doing? What's proven? What's risky? Where are the lessons learned?

Real Federal Deployments

VA (Veterans Affairs): Claims Processing and Medical Records

The VA processes benefit claims, manages medical records, and coordinates care for 9+ million veterans. AI is helping:

• Document classification: Incoming documents (discharge papers, medical records, benefit requests) are classified for routing. A human processes each one, but the AI speeds up sorting.

• Information extraction: Key data (service dates, disability ratings, medical conditions) are extracted from unstructured documents for data entry.

• Scheduling optimization: AI helps optimize appointment scheduling to reduce wait times.

What's working: These are low-risk assistance tasks. The AI speeds up work without making final decisions. Veterans still have human review and appeal processes.

Lessons: VA had to address privacy concerns (protecting veterans' sensitive health data). They work with encrypted systems and only use VA-approved deployments.

IRS: Auditing and Fraud Detection

The IRS receives hundreds of millions of tax documents annually. AI helps:

• Risk scoring: Flagging returns for human review based on patterns (inconsistencies, unusual deductions).

• Document processing: OCR and classification of documents at scale.

• Matching: Identifying mismatches between income reports from employers and taxpayer filings.

What's working: Targets resources where needed most. Humans make all final decisions.

Challenges: Ensuring the system doesn't disproportionately flag certain taxpayer demographics. The IRS has had to audit their system for fairness.

USPS: Mail Sorting and Processing

Post offices process 470+ million mail pieces daily. AI helps:

• Address reading: Optical character recognition reads handwritten and printed addresses.

• Sorting: Routing mail to regions based on address.

• Package identification: Distinguishing letters from packages at scale.

What's working: Clear, repetitive task. High volume. Low individual stakes.

Challenges: Handwriting varies widely. Systems perform worse on certain handwriting styles. USPS continues refining.

EPA: Environmental Monitoring and Satellite Imagery

Environmental agencies monitor air quality, water pollution, and land use. AI helps:

• Satellite image analysis: Detecting illegal dumping, coastal erosion, deforestation.

• Sensor data analysis: Finding patterns in pollution levels.

• Report automation: Extracting key information from environmental reports.

What's working: Enables monitoring at scales humans couldn't achieve. Gives scientists more time for analysis.

Lessons: Satellite imagery varies seasonally and by weather. Systems need context to interpret correctly.

DoD and Security Agencies: Threat Detection (Limited Disclosure)

Military and intelligence agencies are using AI for threat detection, pattern analysis, and logistics. Details are limited for security reasons, but public information shows:

• Pattern recognition in communications and financial data

• Logistics optimization for supply chains

• Image analysis for surveillance

Governance: Heavy restrictions. Limited to specific scenarios. Extensive human oversight. Regular audits for bias.

SSA (Social Security Administration): Claims Processing

The SSA receives millions of retirement, disability, and survivor benefit applications. AI helps:

• Eligibility screening: Flagging applications that clearly meet or don't meet criteria for fast processing.

• Document processing: Extracting information from applications.

• Fraud detection: Identifying suspicious patterns.

What's working: Speeds up processing for clear cases; humans review borderline cases.

Challenges: SSA had to ensure the system didn't disadvantage people with disabilities or certain demographics.

State and Local Examples

Michigan: Unemployment Fraud (Cautionary Tale)

Michigan deployed an AI system to detect unemployment insurance fraud. It flagged 40,000 people as fraud perpetrators. Most were innocent. The system had mathematical flaws. Thousands lost benefits they were entitled to. Lawsuits ensued. Cost: millions in settlements.

Lesson: Don't deploy AI to high-stakes decisions without extensive testing. Don't ignore signs of problems.

California: Benefits Automation

California built an automated benefits system that required citizens to report changes online. The system was poorly designed. Thousands of eligible people lost benefits due to technical failures and confusing instructions. Litigation and negative publicity followed.

Lesson: Automation must account for human diversity. Not everyone is comfortable with online systems. Easy appeals processes are essential.

City of Boston: Hiring and Retention

Boston considered using an AI system to predict which police candidates would be disciplined during their careers. The data showed bias: candidates from certain demographic backgrounds had higher discipline rates—likely because of biased policing practices, not because they were worse police. The city rejected the system.

Lesson: Before deploying AI, audit whether the training data contains the biases you'd be automating.

NYC: Predictive Policing (Discontinued)

New York explored predictive policing: using AI to predict where crime would occur and deploy officers there. The practice amplified over-policing in certain neighborhoods, creating a feedback loop: more police presence led to more arrests, which fed into training data, leading to more predictions, more policing.

Lesson: Prediction systems can amplify existing inequities if not monitored carefully.

Chicago: Police Bias (Cautionary)

Chicago used a risk assessment tool to identify individuals likely to be involved in violence. The tool was biased against African Americans. People on the list faced increased police attention and harassment. The city eventually recognized the bias.

Lesson: High-stakes decisions need fairness audits and human oversight.

International Examples

UK: Post Office Scandal

The UK Post Office deployed a system to detect branch accounting errors. The system was flawed but post office workers were blamed for discrepancies. Hundreds were prosecuted and some imprisoned. Eventually the errors were attributed to the system. The Post Office had to overturn convictions and pay settlements.

Lesson: When an AI system flags problems, assume human error might not be the cause. Investigate the system itself.

EU: GDPR and AI Regulation

The EU is rolling out the AI Act, which classifies AI systems by risk level and requires different governance for each:

• High-risk: systems affecting fundamental rights (criminal justice, benefits, hiring) require extensive documentation, testing, and human oversight.

• Medium-risk: systems like chatbots require transparency disclosures.

• Low-risk: most other systems.

Lesson: Regulatory frameworks are coming. Governance that works will become standard practice.

Anti-Patterns / Misuse Risks

Anti-Pattern 1: Rolling Out Without Testing on Real-World Data

Systems work great in testing but fail in production because real-world data is messier. Lesson: Always test on data matching production conditions.

Anti-Pattern 2: Ignoring Early Signs of Bias

When first deployed, systems often show unequal performance for different groups. Lesson: Monitor from day one. If bias appears, fix it before it harms thousands.

Anti-Pattern 3: Removing Human Review from High-Stakes Decisions

Automating for efficiency but removing the human judgment that should remain. Lesson: AI should assist humans, not replace them in high-stakes decisions.

Practice / Reflection Prompts

• Landscape Mapping: Which federal agencies does your agency interact with? Research their AI deployments. How might it affect you?

• Local Deployment: Does your state or city use AI in any capacity? Research it. What are citizens saying?

• Learn from Failures: Research one government AI failure (Michigan UI fraud, Post Office, etc.). What went wrong? How would you prevent it?

• Peer Learning: Find an agency similar to yours that's using AI. How did they approach governance? What can you learn?

• Readiness Assessment: If your agency deployed AI tomorrow, what would you want from other agencies' experiences?

Key Takeaways

• AI is deployed across federal government in document processing, fraud detection, scheduling, and pattern recognition—mostly assisting humans, not replacing decisions.

• High-stakes AI deployments (Michigan UI, Chicago police) have often caused harm when not carefully tested and monitored for fairness.

• State and local AI often teaches harder lessons about what doesn't work—failures cascade. Learning from them prevents repeating them.

• International regulation is coming (EU AI Act, others)—frameworks that work well are becoming standard practice.

• Successful deployments share characteristics: human oversight, appeals processes, fairness monitoring, and clear governance.

• Failures share characteristics: insufficient testing, ignoring bias, removing human judgment, lack of oversight.

• Your agency is part of this landscape—decisions you make now affect whether future deployments succeed or fail.

Terms / Glossary Items

Production Deployment: When an AI system moves from testing to real-world use affecting real people or processes.

Feedback Loop: When predictions influence the data used to train future systems, potentially amplifying biases.

Fairness Audit: Testing an AI system to see if it performs differently for different demographic groups.

High-Stakes Decisions: Decisions that significantly affect people's lives, rights, or welfare (benefits, criminal justice, hiring for sensitive positions).

Real-World Data: The messy, diverse data the system encounters after deployment, often different from controlled test data.

Equity: Fair treatment and opportunity. AI systems should not systematically disadvantage groups.

Governance: The rules, processes, and oversight mechanisms that guide how an AI system is developed, deployed, and monitored.

You've now seen AI across government: where it works, where it's failed, and the patterns that make the difference. The landscape is your teacher. Learn from others' successes and mistakes.

Your agency is part of this story. When you evaluate AI, you're not just solving your own problems—you're contributing to whether government AI succeeds or fails for everyone.

Next: the policy and legal framework. What rules do you have to follow?

Find one AI deployment in government (federal, state, or local) that affects you or your agency. Research it:

• What does it do?

• Who reviewed it?

• Has anyone complained?

• What would you do differently?

AI in government is real, not hypothetical. The deployments you've read about are affecting people today. Some are working well. Others have harmed people.

Your responsibility: learn from the landscape. Apply lessons to your own context. Help your agency do better.

Next: the policy framework that governs AI use in government.

Government AI CLUB Certification Program

Level 1: AI Aware | What AI Is and Is Not | Lecture 1.1.5

A GOVT.CLUB initiative.

<- 1.1.4 What AI Does Well and Where It Fails 1.1.6 The AI Timeline: Expert Systems to LLMs ->

Start Your CLUB Certification

This lecture is part of L1: AI Aware—8 hours of comprehensive government AI training.

Explore CLUB Certification

Related Lectures

L1 1.1.1—What AI Is and Is Not 20 min - Video + Reading

L1 1.1.2—How AI Actually Works 20 min - Video + Interactive

L1 1.1.3—Types of AI Systems 20 min - Video + Reading

Frequently Asked Questions

What will I learn in AI in Government Today?

In this 20 min video + interactive map lecture, you will Current use cases across federal agencies (VA, IRS, USPS, DoD, EPA, SSA). State and local examples. Interactive map of government AI deployments

What level is AI in Government Today?

This is a Level 1 (AI Aware) lecture, part of Chapter 1.1 \u2014 AI Foundations. It is designed for all government employees.

How long is lecture 1.1.5?

Lecture 1.1.5 (AI in Government Today) takes 20 min. It is delivered as a video + interactive map format.

Do I need prerequisites for AI in Government Today?

This lecture is part of L1 (AI Aware). Prerequisites: None.

What is the CLUB Certification?

CLUB (Community Leading Unified Benchmarks) is a maturity-based AI certification for government professionals with 5 levels (L1-L5), 215 lectures, and 25 chapters aligned with NIST AI RMF, OMB, and GAO frameworks.