Briefing Lawmakers on AI

Learning Objectives

After completing this lecture, you will be able to:

• Understand the key concepts of briefing lawmakers on ai in a government context
• Use downloadable templates for immediate workplace application
• Connect briefing lawmakers on ai to your agency's AI initiatives
• Identify next steps for applying these concepts in your role

Key Topics Covered

-
Congressional testimony preparation

-
Committee briefings

-
One-page policy briefs

-
Translating technical to political

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 cios, chief ai officers, agency technology leaders with the knowledge and frameworks needed to navigate AI in the public sector responsibly and effectively.

As part of the L4 (AI Architect) 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 briefing lawmakers on ai is essential for responsible, effective government AI adoption.

Lecture URL: https://skill.re/learn/govt/briefing-lawmakers-on-ai.php

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

TRANSCRIPT: Briefing Lawmakers on AI

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

You will learn how to explain complex AI concepts to Congressional audiences, anticipate tough questions,

design compelling briefings, and provide Congress with information they need to make good policy. You will

understand what different committees care about, how to structure briefings for impact, and how to handle

political sensitivity around AI governance. By the end, you'll be able to brief Members and staff effectively,

navigate technical-political complexity, and help Congress make informed AI policy decisions.

Congress will ask you about AI. Some Members understand the technology. Most don't. Some have strong views

on AI policy. Others are just beginning to engage. Some care deeply about innovation. Others worry about risk

and rights. Your job is to provide accurate information that Congress can use to make good policy.

This is harder than it sounds. Congress operates on limited time and deep political considerations. You need

to communicate complex technology in plain language. You need to anticipate tough questions about impact and

risk. You need to explain your agency's approach in ways Congress understands and supports. And you need to do

this while Congress is considering competing policy proposals and responding to constituent pressure.

This session walks you through briefing lawmakers on AI. We'll explore how to structure compelling briefings.

We'll discuss how to communicate technical concepts to non-experts. We'll explore how to address political

concerns. And we'll examine how to handle testimony and formal hearings.

COMMUNICATING AI TO CONGRESS EFFECTIVELY

Congress shapes policy. Congress holds the power of appropriations. Congress oversees agency operations. Getting

Congress to understand and support your agency's AI approach is essential.

WHY THIS MATTERS FOR GOVERNMENT

Congressional misunderstanding of AI leads to bad policy. Overly restrictive policies slow innovation. Inadequate

safeguards enable harmful deployments. Policies that don't account for operational constraints are unworkable. Your

job is to help Congress understand the real-world complexity so they can make informed decisions.

UNDERSTANDING YOUR CONGRESSIONAL AUDIENCE

Overview

Different Members and committees have different levels of technical knowledge, different interests, and different

time constraints.

COMMITTEE FOCUS AND INTERESTS

Armed Services Committee / Defense-focused subcommittees

• Interest: National security, defense applications, military advantage, competition with adversaries
• Concern: Is the U.S. maintaining technological edge? What are China and Russia doing?
• Expertise: Low to medium on technology; medium-high on defense implications
• Time constraints: Very limited; focus on strategic implications
• Key message: "We're deploying AI to maintain U.S. military advantage, responsibly"

Intelligence Committee

• Interest: Intelligence collection and analysis, classified AI systems, counterintelligence
• Concern: Are we using AI effectively? Are there security risks?
• Expertise: High on security; low to medium on AI technology
• Time constraints: Very limited; highly classified discussions
• Key message: "AI enhances intelligence effectiveness and security"

Judiciary Committee

• Interest: Rights implications, discrimination, legal liability, regulation
• Concern: Is AI protecting or harming citizen rights? Do we need new laws?
• Expertise: High on law; low on technology
• Time constraints: Moderate; willing to dig into details
• Key message: "We're protecting rights while enabling beneficial AI"

Science Committee

• Interest: Research funding, technical standards, innovation, STEM
• Expertise: Medium to high on science; medium on AI specifically
• Time constraints: Moderate
• Key message: "We're investing in responsible AI research and leadership"

Appropriations Committee / Subcommittees

• Interest: Budget, cost-benefit, return on investment, program effectiveness
• Concern: Are we spending money wisely? What's the return?
• Expertise: Low on technology; high on budgeting
• Time constraints: Very limited; rapid-fire questions
• Key message: "This AI investment delivers value for [beneficiaries]"

Agriculture / Veterans / Homeland Security committees

• Interest: Application to their domain, constituent impact, local implications
• Expertise: High on domain; low on AI
• Time constraints: Moderate
• Key message: "This AI solves real problems in [domain]"

MEMBER TYPOLOGY

Tech Champions

• Know the field; understand technology; advocate for innovation
• Approach: Provide technical depth; discuss innovation opportunities
• Risk: They may push for less oversight than public accepts

Tech Skeptics

• Don't understand technology; skeptical of AI; concerned about risks
• Approach: Explain clearly; address risks directly; show safeguards
• Risk: May restrict beneficial innovation

Pragmatists

• Understand technology reasonably well; want to understand real-world impacts
• Approach: Focus on operations; show what works; acknowledge challenges
• Risk: None significant if you're being honest

Disengaged

• Haven't paid much attention to AI yet
• Approach: Brief them on why it matters; explain implications
• Risk: May default to whatever vocal stakeholders say

STRUCTURING COMPELLING BRIEFINGS

Overview

Good briefing structure works whether you have 5 minutes or 50 minutes.

THE PYRAMID STRUCTURE

Bottom Line Up Front (BLUF)

• Lead with your key point. Don't bury it.
• Examples:
• "The agency is deploying AI systems responsibly and in compliance with federal requirements."
• "This AI system will reduce processing time by 40% while protecting individual rights."
• "We need Congressional approval to expand this capability across government."

Context & Why It Matters

• Provide brief context. Why does Congress care?
• Example: "AI is being used in benefits determination, which affects millions of citizens. Getting this right is

essential for public trust and program effectiveness."

• Be specific about impact: Who is affected? How many people? What's at stake?

Key Facts & Evidence

• Provide 3-5 key facts that support your point
• Use specific numbers: "We've deployed 47 systems... completed 45 impact assessments... identify zero critical risks..."
• Use concrete examples: "In the VA, we use AI to identify veterans at high suicide risk..."
• Use visuals where possible: Charts show impact better than text

Concerns Addressed

• Anticipate tough questions. Address them directly.
• Don't wait to be asked: "You may wonder about bias in this system. Here's what we've found and here's what we're doing..."
• Show you've thought about risks: "Potential concerns: [concern], [concern], [concern]. Here's how we address each..."

Call to Action

• What do you need from Congress?
• Be specific: "We request $X in funding... We need approval for [authority]... We ask for your support..."

TIME MANAGEMENT

5-minute briefing

• BLUF: 30 seconds
• Context: 30 seconds
• Key facts: 2-3 minutes
• Call to action: 30 seconds
• Leave 1 minute for questions

15-minute briefing

• BLUF: 1 minute
• Context: 1 minute
• Key facts: 6-8 minutes
• Concerns addressed: 2-3 minutes
• Call to action: 1-2 minutes
• Leave 1-2 minutes for questions

45-minute hearing testimony

• Prepared statement: 5 minutes (usually read by staff)
• Allows 40 minutes for questions

COMMUNICATING TECHNICAL CONCEPTS TO NON-EXPERTS

Overview

Congress doesn't understand machine learning. You need to explain it in plain language.

ANALOGIES

"Machine Learning is like a translator"

• "You show a machine learning model thousands of examples of English translated to Spanish. It learns patterns.

Then it can translate new sentences it's never seen before. Our AI system learns from historical data the same way."

"Neural Networks are like a brain"

• "Neural networks are inspired by how brains work. They have layers that process information. Early layers learn simple

patterns. Later layers combine those patterns into complex understanding. This is similar to how your brain processes

vision."

"Large Language Models predict the next word"

• "ChatGPT works by predicting the next word in a sequence. You give it a prompt. It predicts the most likely next word.

Then it predicts the next word after that. This continues until it generates a full response. It's very sophisticated

prediction, but that's the core mechanism."

"Bias in AI is like human bias"

• "Just like humans can have biases, AI systems can amplify biases in training data. If historical data shows certain

groups being denied benefits unfairly, the AI learns that pattern. That's why we test for bias carefully."

CONCRETE EXAMPLES (BEATS ABSTRACT CONCEPTS)

Instead of: "The system uses supervised learning with labeled data"

Say: "We trained the system on 10 years of actual benefit determinations. We marked which determinations were correct

and which had errors. The system learned from these examples."

Instead of: "The model achieves 94% accuracy on validation set"

Say: "When we tested the system on cases it hadn't seen before, it made the right decision 94% of the time. This is

better than our human processors (90% accuracy) but we still review all decisions."

Instead of: "We implement adversarial robustness testing"

Say: "We deliberately try to trick the system. We feed it unusual inputs to see if it fails. We test what happens when

data is wrong or adversaries try to manipulate it."

AVOID JARGON

Bad: "We implement XAI techniques using SHAP values to provide attribution explanations"

Good: "We explain which factors influenced the decision. If someone is denied benefits, we can show which factors mattered:

income was over the limit, assets exceeded threshold, work history showed disqualification."

Bad: "The system exhibits hallucinations in generative tasks"

Good: "The system sometimes makes up information. If you ask it to write a report on the Smith family, it might invent

details that aren't true."

Bad: "We conduct red team exercises against frontier models"

Good: "We have security experts try to break the system. They find vulnerabilities we fix. This makes the system stronger."

ADDRESSING POLITICAL CONCERNS

Overview

Different Members care about different things. Anticipate and address their concerns:

FOR CIVIL LIBERTIES FOCUSED MEMBERS

Concern: "This AI system will discriminate against vulnerable populations"

Response structure:

• Acknowledge the concern: "You're right to be concerned. Discrimination is a real risk with AI."
• Show you've addressed it: "We've tested extensively for bias. Results: [specific findings]."
• Show ongoing monitoring: "We monitor for disparate impact quarterly. We have escalation processes if we find issues."
• Show commitment to remedy: "If we discover bias, we pause the system, investigate, and fix it. Here are examples..."
• Invite oversight: "We welcome Congress's oversight. We'll report findings quarterly."

FOR INNOVATION FOCUSED MEMBERS

Concern: "These requirements will slow innovation"

Response structure:

• Acknowledge the concern: "Innovation matters. We want to deploy beneficial AI quickly."
• Show you're not over-restricting: "We have fast-track approval for low-risk systems. Assessment takes weeks, not years."
• Explain why safeguards enable innovation: "Public trust depends on safeguards. When the public trusts AI, we can deploy

more systems."

• Show you're enabling innovation: "We're creating sandboxes for experimentation. We're funding research. We're building

shared infrastructure."

FOR FISCAL HAWKS

Concern: "This will be expensive"

Response structure:

• Show cost-benefit: "Cost is $X million. Benefits: [specific benefits]. ROI: [years to break even]."
• Show efficiency: "We're using shared infrastructure. Cost per agency has decreased by Y%."
• Show alternatives: "Alternative approaches (manual processing, no AI) would cost $X million more."
• Show long-term value: "Initial investment is high. Year 5 costs are lower. Year 10, this is cost-effective."

FOR NATIONAL SECURITY FOCUSED MEMBERS

Concern: "Is supply chain secure? Are we reliant on China for critical components?"

Response structure:

• Acknowledge concern: "Supply chain security is essential. We take this seriously."
• Show vetting: "All vendors pass security review. Key components are sourced domestically/from allies."
• Show monitoring: "We monitor vendors continuously. We have contingency plans if a vendor fails."
• Show resilience: "We don't rely on any single vendor. We have alternate suppliers for critical components."

HANDLING TOUGH QUESTIONS

Overview

Members will ask hard questions. Be prepared.

QUESTION: "How do we know this AI system isn't just another way to discriminate against people?"

Good response:

• Acknowledge: "That's a fair question."
• Show evidence: "We've tested extensively. Here's what we found: [specific findings]. No disparate impact."
• Show process: "We test for bias quarterly. If we find issues, we fix them. Here's the process."
• Invite scrutiny: "We welcome independent audit. [Auditor name] audited us last year. Results: [positive findings]."

QUESTION: "What happens when the AI makes a mistake?"

Good response:

• Be honest: "AI systems make mistakes. This system makes mistakes ~[X%] of the time."
• Show it's better than alternative: "The alternative (human decision-making) has similar error rate, but AI is faster."
• Show oversight: "Every decision is reviewed. We catch most errors before they affect people."
• Show correction: "When errors are discovered, we correct them. Here's an example of a system error we caught and fixed."

QUESTION: "How do we know you're actually following your AI governance policy?"

Good response:

• Show evidence: "We report quarterly to OMB. Reports show: [metrics showing compliance]."
• Show audit: "We've been audited by [auditor]. Results: [findings]."
• Show transparency: "We publish our AI system inventory publicly at [website]. People can see what systems we're deploying."
• Invite challenge: "If you find systems we're not complying with, let us know. Here's how to report concerns: [mechanism]."

QUESTION ABOUT WHAT YOU DON'T KNOW

If asked: "How do you know bias testing is adequate?"

Bad response: "We're confident our testing is adequate."

Good response:

• Be honest: "We don't know for certain. This is a complex problem."
• Explain your approach: "We're using industry-standard testing approaches. We're transparent about limitations."
• Show it's improving: "We're funding research to improve bias testing. We work with universities on this."
• Show humility: "If better testing approaches emerge, we'll adopt them."

TESTIMONY AND FORMAL HEARINGS

Overview

You may be asked to testify before Congress. Prepare carefully.

PREPARING TESTIMONY

Research the Committee

• What bills are they considering?
• What recent hearings have they held on AI?
• What are Members' known positions?
• What did they ask previous witnesses?

Prepare Talking Points

• Your core message in 2-3 sentences
• Supporting points for each
• Examples
• Anticipated questions and answers

Prepare Visual Aids

• Slides you can reference (if time permits)
• Charts showing impact
• Keep visuals simple

Practice

• Practice testimony multiple times
• Have colleagues ask tough questions
• Time yourself (usually 5 minutes for opening statement)
• Refine based on practice feedback

Prepare Written Testimony

• Usually submitted in advance
• Should be comprehensive but scannable
• Include key facts, examples, and points
• Provide exhibits if helpful

DURING TESTIMONY

• Speak clearly and deliberately
• Make eye contact with Members
• Don't read from notes if possible; refer to them, but speak naturally
• Pause for questions
• Don't speculate; stick to what you know
• If you don't know something, say so

ANSWERING QUESTIONS

• Listen to the full question before answering
• Answer the question asked, not the question you wish was asked
• Be concise; a 30-second answer usually suffices
• If you make a mistake, correct it: "Actually, I misstated that. The correct figure is..."
• If you need to follow up with more information, offer to provide it in writing
• Be respectful of Members even if they're adversarial

ANTI-PATTERN 1

Description: Using too much jargon; assuming Congress understands technical details.

Risk: Congress doesn't understand your message. You lose credibility.

What goes wrong: You explain gradient descent optimization. Congress's eyes glaze over. You haven't explained why it matters.

How to avoid: Use analogies. Use examples. Avoid jargon. Test your explanation on someone outside your field.

ANTI-PATTERN 2

Description: Responding to concerns defensively rather than engaging honestly.

Risk: Congress sees you as hiding something. You lose trust.

What goes wrong: Member asks "Are there risks with this system?" You respond "No, it's totally safe."

Congress knows no system is totally safe. They think you're hiding risks.

How to avoid: Acknowledge risks. Explain how you're managing them. Show you've thought about downsides.

ANTI-PATTERN 3

Description: Making claims about system performance, fairness, or safety with more certainty than warranted.

Risk: Reality proves you wrong. Congress loses trust. You lose credibility for future briefings.

What goes wrong: You claim "This system achieves 99% accuracy." It turns out in real-world use, accuracy is 87%.

Congress feels misled.

How to avoid: Be honest about uncertainty. Use ranges: "Accuracy is 85-90% in controlled tests; we're still learning

about real-world performance."

ANTI-PATTERN 4

Description: Not preparing for hard questions Congress will ask.

Risk: When asked, you're unprepared. You give weak answers. You lose credibility.

What goes wrong: You brief on benefits of AI hiring. Congress asks "Doesn't this discriminate against protected classes?"

You don't have a good answer ready.

How to avoid: Prepare for the hardest questions. Identify what Members will care about. Prepare good answers.

PROMPT 1

Select a committee that oversees your agency. Research: What are they interested in? What bills are they considering?

What are Members' known positions on AI? What would they want to hear?

PROMPT 2

Design a 15-minute briefing on your agency's primary AI system for the committee you selected. Include: BLUF, context,

key facts, concerns addressed, call to action.

PROMPT 3

Explain a technical aspect of your AI system (how it works, how you test for bias, how you implement transparency) to

a Member of Congress who knows nothing about AI. Use analogies and examples.

PROMPT 4

Identify 5 tough questions Congress might ask about your AI systems. Develop good answers to each. Have colleague

ask you the questions. Refine your answers.

PROMPT 5

Prepare a 5-minute opening statement for Congressional hearing on your agency's AI initiatives. Practice it. Have

colleague critique it. Refine it.

• Different committees and Members care about different things. Know your audience.
• Lead with your bottom line. Structure briefings for clarity and impact.
• Use analogies and examples. Avoid jargon. Assume Congress doesn't understand technology.
• Anticipate tough questions and address them directly. Show you've thought about risks and concerns.
• Be honest about uncertainty. Don't overstate confidence in system performance or safety.
• When testifying, be prepared, speak clearly, and respect Members even if they're adversarial.
• Build relationships over time. One briefing won't change minds. Consistent, honest communication does.

BLUF (BOTTOM LINE UP FRONT)

Key point or conclusion stated at the beginning of briefing. Ensures audience gets main message even if they

don't listen to entire briefing.

MARK-UP

Committee process where bill language is debated and amended before vote.

FLOOR VOTE

Voting process where full chamber votes on legislation.

APPROPRIATIONS

Congressional action allocating funding to specific programs or agencies.

OVERSIGHT

Congressional responsibility to monitor agency operations and compliance with law.

Congressional briefing is both art and science. You need to understand your audience, communicate clearly, anticipate

concerns, and build credibility over time. Get these things right, and Congress becomes a partner in developing AI

policy. Get them wrong, and Congress becomes a barrier to effective AI governance.

The agencies that excel at Congressional engagement treat it seriously. They invest in people with communication skills.

They prepare carefully. They engage honestly. They build relationships. They educate Congress so Congress can make good policy.

Take 20 minutes to identify:

• What committees oversee your agency on AI matters?
• What do those committees care about most?
• What tough questions will they ask about your systems?
• How prepared are you to brief them effectively?
• What would improve your Congressional engagement?

Congress will make AI policy. Your job is to help Congress understand the real-world complexity so they can make

informed decisions. Do this well, and the policies Congress creates will be better. Do it poorly, and the policies

will constrain your agency needlessly. Invest in this capability. It matters.

Government AI CLUB Certification Program

Level 4: AI Architect | Briefing Lawmakers on AI | Lecture 4.2.7

A GOVT.CLUB initiative.

<- 4.2.6 Public Consultation on AI Policy
4.2.8 Privacy Policy Engineering ->

Start Your CLUB Certification

This lecture is part of L4: AI Architect -- 120 hours of comprehensive government AI training.

Explore CLUB Certification

Related Lectures

L4
4.2.1 -- Drafting Agency AI Policies
180 min - Workshop + Template

L4
4.2.2 -- Navigating the Federal AI Landscape
150 min - Lecture + Tracker

L4
4.2.3 -- State and Local Government AI Policy
150 min - Lecture + Comparison