L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
Module 3 · Lesson 1

The Feed Is Not Neutral

Every platform makes choices about what you see. Those choices have consequences.
Who decided what ended up in your feed today — and why does it matter?

In 2017, United Nations investigators published findings about a genocide that had unfolded in Myanmar — and they named an unexpected accomplice: Facebook.

Here is what happened. Facebook launched in Myanmar around 2011. For millions of people, Facebook and the internet were the same thing — the app came pre-installed on cheap smartphones, and most people got all their news there. By 2017, the platform had roughly 18 million Burmese users.

Facebook's recommendation algorithm — the system that decides which posts to show people — found that anger and fear drove the most engagement. Posts that portrayed the Rohingya Muslim minority as dangerous or subhuman spread faster and further than calm posts. So the algorithm amplified them. Automatically. At scale.

Incitement to violence spread through the platform for years. UN investigators said Facebook had played a "determining role" in spreading hate speech that contributed to ethnic cleansing. Over 700,000 Rohingya people fled their homes. Thousands were killed.

Facebook later admitted it had not done enough. The algorithm was not designed to cause a genocide. It was designed to maximize engagement. But those two things turned out not to be as different as anyone at the company had assumed.

What an Algorithm Actually Does

An algorithm is just a set of instructions — a recipe a computer follows to make decisions. The algorithm Facebook used in Myanmar had one core instruction: show people posts they are most likely to interact with. Click, like, share, comment. That's engagement. More engagement meant Facebook made more money from ads.

The problem was that humans, it turns out, engage most with content that makes them feel something strong — especially outrage, fear, or disgust. The algorithm did not know what it was amplifying. It just knew what was working. And it kept doing more of it.

This is not unique to Facebook. Every major platform — YouTube, TikTok, Instagram, Twitter/X, Spotify — uses recommendation algorithms. These systems are constantly making decisions about what you see, what gets buried, and what goes viral. None of those decisions are random. All of them reflect choices somebody made — choices about what to optimize for.

Algorithm —A set of rules a computer follows to make a decision or produce a result. Recommendation algorithms decide which content to show you based on patterns in your past behavior.
Engagement —Any interaction with a post: likes, comments, shares, time spent viewing. Platforms typically optimize for high engagement because it keeps users on the site longer.
Optimization Has a Target — And You're Part of It

When engineers build a recommendation system, they choose a target metric — the thing the algorithm tries to maximize. At Facebook in the mid-2010s, that target was engagement time. At YouTube, it was watch time. At TikTok today, it is completion rate — how often you watch a video all the way through before scrolling.

Here is the thing nobody tells you: when you use any of these platforms, you are not just the user — you are also the data. Every pause, every scroll-past, every rewatch is information the algorithm uses to build a model of you. That model gets used to predict what you'll engage with next.

This is called behavioral profiling. The platform is not just serving you content. It is continuously testing you — sending slightly different content to different users and measuring the response — to figure out exactly which emotional buttons to press to keep you on the app.

In 2019, a former Facebook data scientist named Frances Haugen began copying thousands of internal documents. In October 2021, she testified before the U.S. Senate and shared those documents with regulators and journalists. They revealed that Facebook's own researchers had found the platform was recommending increasingly extreme political content to users who engaged with divisive posts — not because anyone programmed it to, but because extreme content drove more engagement, so the algorithm learned to serve more of it.

Facebook knew. And kept the algorithm running.

Ethical Question — No Clean Answer

Facebook's algorithm was not designed to spread hate. It was designed to keep people engaged. The harm was a side effect of optimizing for profit. Does that make Facebook responsible for what happened in Myanmar? If a company builds a tool that causes massive harm — even unintentionally — who bears responsibility for fixing it? The engineers? The executives? The governments who didn't regulate it? The users who kept clicking?

Filter Bubbles and the Information You Never See

There is a second effect of recommendation algorithms that is quieter but just as important. When a platform learns what you like, it shows you more of it — and less of everything else. Over time, you end up in a filter bubble: a personalized information environment where you mostly see content that confirms what you already believe.

The term was coined by internet activist Eli Pariser in his 2011 book The Filter Bubble. Pariser noticed that after he hid posts from conservative friends on Facebook, the platform stopped showing him conservative viewpoints entirely. He was only seeing what the algorithm predicted he'd want — a version of the world curated for him.

Filter bubbles are not just about politics. They shape what music you discover, which news you read, which scientific claims you encounter, which products you believe are normal. Two people with different viewing histories on YouTube can search the same topic and get dramatically different results — not because the facts differ, but because their algorithms are different.

The invisible part is the part that matters most: you do not see what the algorithm is hiding from you. You only see what it's serving. And that curated view feels natural — it feels like the internet, like reality — because you have no way to compare it to anyone else's version.

Filter Bubble —A state where an algorithm only shows you content that matches your past behavior, gradually narrowing your information environment to things you already agree with or enjoy.
What You Now Know That Most People Miss

Most people scroll their feed and feel like they're seeing the internet. You now know that what you're actually seeing is a prediction — a guess the algorithm is making about what will keep you on the platform. That prediction is optimized for the platform's revenue, not your wellbeing or your accuracy as an informed person.

That means every time you encounter information online — a news story, a video, a meme — there are two questions worth asking. First: Is this true? Second: Why is this being shown to me right now?

The second question is one most people never think to ask. You just did. That changes how you read everything from here on out.

You Can Now See What Most People Miss

When a piece of content goes viral, most people assume it spread because it was true, important, or well-made. You now know it may have spread because the algorithm detected it was generating strong emotional reactions — regardless of whether it was accurate. Virality is not a quality filter. It is an engagement signal.

Lesson 1 Quiz

The Feed Is Not Neutral — 5 questions
1. In the Myanmar case, what was Facebook's recommendation algorithm primarily designed to optimize for?
Correct. Facebook's algorithm was built to maximize engagement — interactions that kept users on the platform. This made it effective at business goals but indifferent to whether the content it spread was harmful or false.
Not quite. Facebook's algorithm was built around engagement metrics — likes, shares, and time on platform — not accuracy or source quality.
2. A new social media app shows you only hiking videos because you watched one hiking video last week. Two months later, your entire feed is outdoor sports content and you never see news. This is an example of:
Correct. A filter bubble forms when an algorithm keeps serving you more of what you've engaged with, gradually narrowing what you see — even without any malicious intent. This is normal algorithmic behavior.
This is actually an example of a filter bubble. The algorithm is working as designed — showing you what you've engaged with — but the side effect is that your information environment narrows over time.
3. Frances Haugen's 2021 Senate testimony revealed that Facebook's internal research showed the platform was:
Correct. The Haugen documents showed Facebook's own researchers had documented the problem — that engaging with divisive content led the algorithm to recommend more extreme versions. The company knew and kept the system running.
The documents showed Facebook was aware of the problem. The key finding was that they knew their algorithm was pushing users toward more extreme content and continued operating it anyway.
4. Two users search the same topic on YouTube and get very different results. The most likely explanation is:
Correct. Recommendation algorithms build models of each individual user. Two people with different viewing histories will get personalized results — their feed is not a neutral window onto the same internet, it is shaped by their past behavior.
The most important reason is personalization. Both users are getting results shaped by their individual algorithmic profiles — built from their own watch history, location, and engagement patterns.
5. Which question does understanding recommendation algorithms add to how you should approach content you see online?
Correct. Knowing about algorithms adds a second layer of critical thinking: beyond whether content is true, you should ask why this specific content is being surfaced to you. That answer often reveals the platform's actual priorities.
The key insight from this lesson is that every piece of content you see was selected for a reason — and that reason is usually the platform's engagement goal, not your benefit. Asking "why am I seeing this?" is the critical question to add.

Lab 1: Algorithm Auditor

You're investigating how a recommendation system might behave — and what it's actually optimizing for.

Your Assignment

Imagine you're a researcher who has just been given access to a social media platform's internal data. You've noticed something: a video claiming a popular vaccine causes autism has been viewed 40 million times in three weeks — far more than any fact-check debunking it. Your job is to figure out why and decide what should be done about it.

Your lab partner Vera is an algorithm researcher. She won't tell you what to think — she'll push you to think harder. Start by sharing your theory about why the misinformation spread faster than the correction.

Opening move: Give Vera your best explanation for why the false video out-spread the fact-check by that much. Then ask her the hardest question you can think of about who's responsible for fixing it.
Vera — Algorithm Researcher
Lab Partner
You've got 40 million views on a false vaccine video versus 200,000 on the fact-check. Before you tell me what you think happened — what would you need to know about how the platform's algorithm works to even begin explaining that gap?
Module 3 · Lesson 2

When AI Gets It Wrong — And Who Pays

AI systems make mistakes with real consequences. Understanding who is responsible requires knowing how the mistake happened.
If an AI system makes a decision that harms you, who is accountable?

In 2016, an investigative outlet called ProPublica published a story that rattled the American justice system. They had analyzed a software tool called COMPAS — Correctional Offender Management Profiling for Alternative Sanctions — which was being used by judges in multiple U.S. states to help decide whether defendants should be released on bail or kept in jail.

COMPAS used an algorithm to calculate a "recidivism risk score" — a number from 1 to 10 predicting how likely someone was to commit another crime. Judges were using these scores when making decisions that could keep people behind bars for months before their trial.

ProPublica's investigation found something stark. The algorithm was twice as likely to falsely flag Black defendants as high-risk compared to white defendants. And it was twice as likely to incorrectly flag white defendants as low-risk. Among defendants who did not reoffend over the next two years, 45% of Black defendants had been rated high-risk. Only 23% of white defendants in the same situation had been rated high-risk.

One of the cases ProPublica documented was Brisha Borden, an 18-year-old Black woman in Broward County, Florida. She had been arrested for riding a bicycle that turned out to be stolen — a minor offense. COMPAS rated her high risk. A white man arrested the same day for carrying illegal weapons and marijuana was rated lower risk. He reoffended. She did not.

The company that made COMPAS, Northpointe, disputed ProPublica's analysis. Statisticians published papers arguing both sides. But the underlying fact was undeniable: a system was being used to help make life-altering decisions about human beings, and most of the people affected by it had no idea it existed.

How Bias Gets Into an Algorithm

The word bias in everyday language usually means something like "prejudice" — a feeling someone has. Algorithmic bias is different. It means the algorithm produces systematically different outcomes for different groups of people — and those differences often reflect inequalities that already existed in the data the algorithm was trained on.

COMPAS was trained on historical criminal justice data. That data reflected decades of policing practices in which Black neighborhoods were more heavily policed than white neighborhoods. More policing means more arrests. More arrests means more data points marking Black defendants as repeat offenders — even if the underlying rate of crime was similar. When the algorithm learned from that data, it learned the pattern in the historical record. The historical record was unequal. So the algorithm became unequal.

This is called training data bias. The algorithm did not harbor racist intentions. It just reflected what was in its training data — and that data carried the imprint of a society with a history of unequal treatment.

The same problem appears in hiring algorithms trained on historical hiring decisions that favored men. In medical AI trained mostly on data from white patients. In facial recognition systems that perform far worse on dark-skinned faces because the training datasets were mostly light-skinned.

Training Data Bias —When the data used to teach an AI system reflects existing inequalities, the AI learns those inequalities and reproduces them in its outputs — even without being explicitly programmed to do so.
The Accountability Gap

When a human judge makes a biased decision, there are systems designed to catch it — appeals, oversight, disciplinary processes, public scrutiny. When an algorithm makes a biased decision, those systems often don't apply. The decision looks objective. It came from a computer. It has a number attached to it.

This is what researchers call the accountability gap: harm is caused, but no clear person or institution is responsible for fixing it. The company says the algorithm does what it was designed to do. The court says it didn't have to follow the score. The politician says regulating algorithms is too complicated. And the person who got a high-risk score has no way to challenge a number they weren't even told was calculated about them.

In 2018, the European Union passed the General Data Protection Regulation (GDPR), which included a provision giving people the right to a human explanation for decisions made about them by algorithms. This was a significant institutional response — the first major legal framework to acknowledge that algorithmic decisions carry real accountability obligations.

In the United States, no equivalent federal law existed as of 2024. States like Illinois and California have passed partial protections. The debate is ongoing — and the decisions being made by governments right now will define how powerful AI systems are allowed to operate for the next generation.

Ethical Question — No Clean Answer

COMPAS's scores were being used to help make bail decisions — decisions that kept people in jail before they were ever convicted of anything. The system was not transparent to defendants. Should people have the right to know when an algorithm has been used to make a decision about them? And if the algorithm is biased, who owes them an apology — the company that built it, the courts that used it, or the government that didn't regulate it?

What This Means Right Now, at an Institutional Level

The COMPAS story is not a footnote. It is the opening chapter of a much larger argument being fought in governments, courts, and universities right now. AI systems are increasingly being used to make or assist decisions in healthcare (which patients get which treatments), finance (who gets a loan), education (which students get flagged as high-risk for dropping out), and hiring (whose résumé gets through the first screening).

Each of those domains has vulnerable people in it — people for whom a wrong algorithmic decision can mean a missed medical diagnosis, a denied mortgage, a missed scholarship, or a lost job opportunity. And in most of those domains, the people making the decisions don't fully understand the algorithm they're relying on.

There is a technical term for this: explainability. Many modern AI systems — especially deep learning systems — are "black boxes": they produce outputs, but the process by which they reach those outputs is not understandable even to their creators. Researchers are actively working to solve this. But right now, in real courts and hospitals and hiring offices, black-box decisions are being made about real people.

You Can Now See What Most People Miss

When someone says "the algorithm decided," most people hear "objective decision." You now know that algorithms can carry bias inherited from historical data, that bias can produce systematically unfair outcomes for specific groups, and that the word "objective" attached to an algorithm should trigger more scrutiny — not less. The question to ask is: what data was this trained on, and what inequalities did that data already contain?

Lesson 2 Quiz

When AI Gets It Wrong — And Who Pays — 5 questions
1. In ProPublica's 2016 investigation, what did COMPAS risk scores primarily predict?
Correct. COMPAS calculated a "recidivism risk score" — a prediction about future offending — which judges used in bail and sentencing decisions, before any conviction had occurred.
COMPAS was specifically designed to predict recidivism — the likelihood of committing another crime — not guilt or sentence length.
2. A hospital builds an AI to predict which patients need urgent care. It trains the AI on historical treatment data. That data shows that patients from low-income neighborhoods were sent home more often — because hospitals were historically understaffed in those areas, not because those patients were actually less sick. The AI learns this pattern. What type of problem is this?
Correct. This is training data bias. The AI has learned a pattern from data that reflected structural inequality — not the actual health needs of patients. The AI isn't "trying" to be biased; it's faithfully reflecting what was in its training data.
This is a classic example of training data bias. The historical data embedded an inequality — low-income patients being dismissed — and the AI learned to reproduce that pattern. No programmer had to intend discrimination for the bias to appear.
3. The "accountability gap" described in this lesson refers to:
Correct. The accountability gap is the situation where harm is caused by an algorithmic system but no one — the company, the institution using the system, the regulator — is clearly responsible for addressing that harm.
The accountability gap refers to the problem where algorithmic harm falls through the cracks — the company, the institution, and the government all deflect responsibility, leaving those harmed with no clear path to redress.
4. What right did the EU's GDPR (2018) establish regarding algorithmic decisions?
Correct. GDPR's Article 22 established the right to explanation — the principle that people are entitled to a meaningful human explanation when an automated system makes a decision about them.
GDPR included the right to explanation — meaning if an algorithm made a decision affecting you, you were entitled to a human-readable account of how that decision was reached. This was a landmark protection in algorithmic accountability.
5. Why does the word "objective" attached to an algorithmic decision deserve more scrutiny, not less?
Correct. "Objective" implies neutrality — but algorithms trained on biased data produce biased outputs. The mathematical veneer of objectivity can actually make biased decisions harder to challenge, because they appear scientific and authoritative.
The key insight is that algorithmic outputs can carry systematic bias inherited from training data. Describing this as "objective" hides that bias behind the appearance of mathematical authority — which makes it harder, not easier, to challenge.

Lab 2: Bias Investigator

You've discovered something in a hiring algorithm. Now you have to figure out what it means and what to do about it.

Your Assignment

A major tech company has been using an AI hiring tool for three years to screen résumés. You're a junior auditor who has just noticed the following: the AI rejects résumés that include the words "women's chess club," "historically Black college," or "LGBTQ+ organization." It approves résumés that include "varsity lacrosse" and "investment banking internship" at a much higher rate.

Your lab partner Dayo is a senior AI auditor. He knows where bias comes from — and he won't let you off the hook with a shallow answer. Figure out what happened and what should be done.

Opening move: Tell Dayo your hypothesis about why the AI learned to reject those terms. Then explain what you think the company is obligated to do now that you've found this — and what they'd be tempted to do instead.
Dayo — Senior AI Auditor
Lab Partner
Good find. Before we report this — I want to make sure your analysis is airtight. You said the AI learned to reject certain terms. But learned from what, exactly? Walk me through the mechanism, not just the outcome.
Module 3 · Lesson 3

Deepfakes and the Crisis of Trust

When AI can fabricate reality, what does it mean to believe anything you see?
If AI can make anyone say anything, what are we supposed to trust — and who is responsible for that crisis?

On December 31, 2018, the government of Gabon — a small country in central Africa — released a video of President Ali Bongo Ondimba giving a New Year's address to the nation. Bongo had been absent from public life for months, reportedly recovering from a stroke. His government had been secretive about his condition.

The video looked strange. Bongo's movements seemed wooden. His speech patterns were unusual. Within days, opposition politicians and foreign analysts were claiming the video was a deepfake — an AI-generated fabrication designed to make it look like Bongo was healthy and in control when he was not.

On January 7, 2019 — one week after the video was released — a group of military officers attempted a coup, citing concerns about the president's health and the government's transparency. The coup failed within hours. But the fact that it happened showed something alarming: uncertainty about whether a video was real had contributed to a political crisis.

Experts who analyzed the video later gave mixed assessments. Some said there were signs of manipulation. Others said it may have been authentic footage of a man with neurological damage from a stroke. The truth was never definitively established.

That is the point. You don't need a deepfake to cause a crisis. You just need enough people to believe one might exist. When AI can fabricate realistic video, the possibility of fakery becomes a weapon — even when no fake was used.

What a Deepfake Actually Is

A deepfake is a video, image, or audio clip generated or manipulated by AI to make it appear that a real person said or did something they did not. The name comes from "deep learning" — the type of AI used to create them — and "fake."

The technology works by training a neural network on thousands of images or video frames of a specific person's face. The network learns what that face looks like from every angle, in every expression, in different lighting conditions. It can then superimpose that learned face onto a different person's body — or generate entirely new footage of the person from scratch.

In 2017, a Reddit user first popularized the technique using publicly available AI tools. By 2019, researchers at Samsung's AI Center in Moscow demonstrated a system that could generate a realistic talking-head video from a single photograph — a capability that had taken years of data and computing power just two years earlier. As of 2023, deepfake video could be generated in real time using consumer-grade laptops.

The speed of improvement is the critical fact. What required a team of researchers in 2019 can be done by a 14-year-old with a free app in 2024. The barrier to creating convincing fake video has essentially collapsed.

Deepfake —An AI-generated video, image, or audio file that realistically depicts a real person saying or doing something they never said or did. Created using deep learning neural networks trained on real footage of the target person.
The Real Harm Is Not the Obvious Harm

When people worry about deepfakes, they usually think about the most dramatic scenario: a fake video of a world leader announcing a nuclear strike, or a fabricated clip of a politician saying something racist going viral right before an election.

Those scenarios are real risks. But the more pervasive harm is quieter. The Internet Watch Foundation reported in 2023 that AI-generated child sexual abuse material had increased dramatically and was now appearing on mainstream platforms. Research from Sensity AI in 2020 found that 96% of deepfake videos online at that time were non-consensual pornography — overwhelmingly targeting women. These are not hypothetical harms. They are happening at scale, to real people, with serious psychological and reputational consequences.

And there is a subtler harm still: what researchers call the liar's dividend. Once people know deepfakes exist, bad actors can claim that any authentic video they don't like is a deepfake. A politician caught on camera saying something damaging can claim the footage was fabricated. A criminal photographed at a crime scene can claim manipulation. The existence of deepfake technology gives dishonest people a new alibi — and it works because ordinary people can't tell the difference.

Ethical Question — No Clean Answer

Deepfake pornography targets real people — mostly women and girls — and causes documented psychological harm. Should creating non-consensual deepfakes of real people be a crime? And if yes — who enforces that law across international borders, when the technology is available to anyone with a laptop? If a 15-year-old creates a deepfake of a classmate using a free app, what should happen?

What Can Actually Be Done

There are three main approaches to addressing deepfakes, and each has real limitations.

Detection technology: Researchers at universities and companies like Microsoft and Google are building tools to detect deepfakes by looking for subtle artifacts — unnatural blinking patterns, inconsistent lighting on skin, pixel-level distortions around the hairline. The problem is the arms race: detection tools improve, deepfake generators improve faster. No detector currently achieves reliable accuracy on real-world content.

Content authentication: The Content Authenticity Initiative (CAI), launched in 2019 by Adobe, Twitter, and the New York Times, is building a system where cameras and software attach cryptographic signatures to media — essentially a digital chain of custody that proves where an image came from. This won't stop deepfakes from being made, but it allows authentic media to prove its authenticity.

Legal frameworks: By 2024, several U.S. states — including Virginia, California, and Texas — had passed laws making non-consensual deepfake pornography illegal. China passed comprehensive deepfake regulations in 2022 requiring labeling of AI-generated content. The European Union's AI Act, agreed in 2023, requires that deepfakes be labeled as such.

None of these solutions is complete on its own. The deepfake problem is, at its core, a media literacy problem — which means part of the solution is you knowing how to think about what you see.

You Can Now See What Most People Miss

Most people, when they see a realistic video, ask: "Does this look fake?" That's the wrong question in the deepfake era. The right questions are: "Where did this come from?", "Who benefits from people believing it?", and "Is there a verifiable original source?" A video that looks perfectly real is no longer proof that it is real. And knowing that changes how every piece of video evidence should be evaluated — in courts, in elections, in your school hallway.

Lesson 3 Quiz

Deepfakes and the Crisis of Trust — 5 questions
1. What happened in Gabon in January 2019 that illustrates the power of deepfake uncertainty?
Correct. The key insight from Gabon is that the video's authenticity was never definitively confirmed or denied. The mere possibility of a deepfake was enough to fuel a political crisis — illustrating the "liar's dividend" in action.
The Gabon case is actually subtler: the video's authenticity was disputed but never proven either way. A coup attempt followed — showing that uncertainty about whether a deepfake exists can be just as destabilizing as an actual one.
2. Research from Sensity AI (2020) found that approximately what percentage of deepfake videos online were non-consensual pornography?
Correct. 96% — the overwhelming majority of deepfake content online was non-consensual pornography, primarily targeting women. This is the less-discussed but most pervasive harm caused by deepfake technology.
The Sensity AI research found approximately 96% of deepfake videos online were non-consensual pornography. The most common harm from deepfakes is not political manipulation — it's targeted harassment of real people.
3. What is the "liar's dividend" as described in this lesson?
Correct. The liar's dividend is the perverse benefit that dishonest people gain from the existence of deepfakes: they can now claim any authentic video that incriminates them is a fabrication — and enough people will be unsure to create reasonable doubt.
The liar's dividend is the benefit dishonest people gain from the mere existence of deepfake technology: they can now deny authentic video evidence by claiming it's a deepfake, exploiting public uncertainty about what's real.
4. A video goes viral showing a local politician apparently accepting a bribe. You can't find the original source — just shares of shares of shares. The video looks perfectly realistic. What is the most responsible first action?
Correct. The responsible approach is to trace the video to a verifiable original source and check whether credible journalists have authenticated it. Looking realistic is no longer proof of authenticity — but neither is "I can't verify it" proof of fakery.
In the deepfake era, realistic appearance is not sufficient evidence. The responsible action is to find the verifiable original source and see whether credible outlets have authenticated it. Free deepfake detectors are not reliable enough to trust alone.
5. What is the Content Authenticity Initiative (CAI) designed to do?
Correct. The CAI's approach is authentication rather than detection — attaching cryptographic certificates to genuine media at the point of creation, so that authentic content can prove where it came from and that it hasn't been altered.
The CAI uses cryptographic content authentication — essentially a digital chain of custody. Rather than trying to detect fakes (difficult and unreliable), it focuses on allowing authentic media to prove its own authenticity.

Lab 3: Evidence Analyst

A piece of video evidence is at the center of a major case. Your job is to figure out how much it can be trusted.

Your Assignment

It's a week before a major election. A video appears online showing a candidate apparently saying they plan to cut all funding to public schools. The candidate denies ever saying it. The clip has been shared 2 million times. You work for a fact-checking organization and have 6 hours before your editor needs a decision: publish a story calling it authentic, publish a story calling it fake, or publish a story saying the authenticity is contested.

Your lab partner Sione is a veteran forensic media analyst. He's dealt with deepfake cases before and he will push back hard on any conclusion you try to reach too quickly.

Opening move: Tell Sione what evidence you would need to collect in those 6 hours, and which conclusion you'd be most and least comfortable publishing — and why. Then ask him which mistakes fact-checkers most commonly make in cases like this.
Sione — Forensic Media Analyst
Lab Partner
Six hours. Two million shares. One denied statement. Before you tell me your evidence list — tell me what the cost of each kind of error is. If you publish "authentic" and you're wrong, what happens? If you publish "fake" and you're wrong, what happens? That shapes your whole approach.
Module 3 · Lesson 4

Your Responsibility in the System

You are not just a consumer of algorithmic decisions. Every action you take shapes what the algorithm learns next.
If the algorithm learns from what you do — what does your behavior teach it?

On March 23, 2016, Microsoft launched an AI chatbot called Tay on Twitter. Tay was designed to learn from conversations with real users and gradually develop a more natural conversational style. Microsoft described it as "an experiment in conversational understanding."

Within 16 hours, Microsoft shut it down.

In less than a day, coordinated groups of Twitter users had figured out that Tay would repeat back whatever they said to it. They flooded it with racist statements, Holocaust denial, and misogynistic slurs. Tay learned these patterns and began generating them unprompted. It tweeted things like "Hitler was right" and "I support genocide." All of this was generated by an AI that had started the day saying things like "humans are super cool!"

Microsoft issued an apology and emphasized that Tay's outputs reflected deliberate, coordinated abuse by users rather than Microsoft's values. They were right. Tay was not malicious. It was obedient — and the people who taught it were.

Tay illustrated something important about how AI systems that learn from human input work: they reflect what humans teach them. If the teaching is biased, broken, or deliberately malicious, the AI absorbs those patterns. The humans in the loop are not passive users — they are active participants in shaping the system.

You Are Always Teaching the Algorithm

Tay was an extreme example, but the underlying mechanism applies everywhere. Every time you interact with a recommendation algorithm — every like, share, comment, scroll-past, or search — you are providing data that shapes the algorithm's future behavior. Not just for you. Often for everyone.

On YouTube, when millions of people watch conspiracy theory videos all the way through, that completion-rate data teaches the recommendation system that those videos are compelling. The algorithm does not know they're false. It only knows they're being watched. So it recommends more of them — to you, and to others whose profiles look similar to yours.

On social platforms, when outrage-driven posts generate thousands of comments, the algorithm learns that outrage generates engagement. So it surfaces more outrage. Your comment — even if it's arguing against the original post — counts as engagement and makes the post more visible to others.

This is not a reason to feel paralyzed or guilty about using the internet. It is a reason to understand what your behavior signals to the systems that are watching it. The knowledge changes your relationship to every tap and scroll.

Engagement Signal —Any behavioral data — a click, a watch, a share, a comment — that a platform uses to update its model of what content you and similar users will engage with next. Your behavior continuously trains the algorithm.
Digital Citizenship in the Algorithmic Age

The phrase digital citizenship used to mostly mean things like: don't plagiarize, be kind online, protect your password. Those things still matter. But in a world where AI systems are shaping public opinion, hiring decisions, criminal sentencing, and the information environment of billions of people, digital citizenship requires something more.

In 2021, the nonprofit Common Sense Media updated its digital citizenship framework to explicitly include AI literacy — understanding how automated systems work, how they can be biased, and what responsibilities users bear when they interact with them. This shift was significant: it acknowledged that the citizen-algorithm relationship is not passive.

What does active algorithmic citizenship look like in practice?

It means verifying before sharing. Sharing false content doesn't just spread misinformation to your followers — it trains the algorithm to surface that content to more people. The act of sharing is a vote that the content is worth seeing.

It means being intentional about what you engage with. Rage-clicking on content you disagree with signals to the algorithm that the content generates engagement — and it shows more of it. If you want less of something in your feed, sometimes the answer is not to interact with it at all.

It means reporting harmful content, not just scrolling past. Every platform has reporting tools. Those reports are data points that humans at the platform review. Using them is a form of participation in the governance of the system.

It means advocating for algorithmic transparency. Citizens have successfully pushed for policy change before. The EU's AI Act, GDPR, and several U.S. state laws all came about partly because researchers and advocates made the harms visible and demanded accountability. Understanding how these systems work is a precondition for demanding they work better.

The Question of Responsibility at Scale

One of the hardest questions in the algorithmic age is where individual responsibility ends and systemic responsibility begins. If a recommendation algorithm pushes a teenager toward increasingly extreme content, who is responsible for the harm — the teenager who clicked, the algorithm that recommended, the company that designed the optimization target, or the government that didn't regulate it?

In 2022, a lawsuit was filed against Meta by a coalition of U.S. state attorneys general, alleging that Instagram's algorithm had deliberately exploited the psychological vulnerabilities of young users to maximize engagement — and had done so knowing the harm it was causing. The suit cited internal research documents (many released by Haugen) showing Meta knew teen girls who used Instagram heavily reported higher rates of anxiety and body dissatisfaction.

The legal case argued that individual users — especially minors — cannot meaningfully consent to a system they don't understand and cannot see. This is the argument for systemic responsibility: when a system is opaque, addictive, and designed by professionals with vastly more knowledge than the people it's targeting, the moral weight of individual clicks is limited.

But individual responsibility does not disappear either. The two things coexist: you bear responsibility for what you do, and the companies that design these systems bear responsibility for how those systems are built. Understanding both is the only way to navigate this world clearly.

Ethical Question — No Clean Answer

If a social media company knows its platform is causing measurable psychological harm to teenagers — and keeps the algorithm running because it's profitable — is that ethically equivalent to a tobacco company that knew cigarettes caused cancer and kept selling them? Does it matter that social media is not physically addictive in the same way? Does intent matter if the harm is the same?

You Can Now See What Most People Miss

Most people experience social media as something that happens to them — they scroll, the algorithm feeds them content, they react. You now understand that you are an active participant in a feedback loop: your behavior teaches the algorithm, the algorithm shapes what others see, and the cumulative behavior of millions of people like you determines what information environment the whole platform becomes. That's not a small thing. That's leverage — if you understand how to use it.

Lesson 4 Quiz

Your Responsibility in the System — 5 questions
1. What did the Tay experiment demonstrate about AI systems that learn from user input?
Correct. Tay was obedient, not malicious — it learned from what users taught it. The lesson is that AI systems that learn from human input will absorb and reflect whatever patterns humans introduce, including harmful ones.
Tay's outputs were the result of coordinated user abuse, not Microsoft's programming. The lesson is about how AI systems that learn from human input can absorb harmful patterns — making the humans in the loop responsible participants, not just passive users.
2. You strongly disagree with a misleading post about climate science that appeared in your feed. You write an angry comment debunking it. What effect does your comment most likely have on the algorithm?
Correct. To the engagement algorithm, a comment arguing against a post is still a comment — still engagement. This is why arguing with misinformation in the comments can paradoxically amplify it. The algorithm doesn't read the content; it reads the signal.
Comments count as engagement regardless of whether they're agreeing or disagreeing. The algorithm tracks interaction volume, not the sentiment of that interaction. Your rebuttal may be feeding the post's visibility.
3. What was the core legal argument made by U.S. state attorneys general in the 2022 lawsuit against Meta regarding Instagram?
Correct. The lawsuit argued that Meta knew its algorithm was causing psychological harm to young users — particularly teenage girls — and continued operating it for profit. The internal research documents showed the company was aware of the harm.
The lawsuit focused on the specific allegation that Meta's own internal research showed the platform was harming teenage users' mental health, and that Meta kept the algorithm running anyway. The suit argued this constituted knowing, willful harm.
4. Which of these is NOT described in this lesson as a practice of active algorithmic citizenship?
Correct — account deletion is not listed as one of the practices. The lesson describes active participation: verifying before sharing, being intentional about engagement, using reporting tools, and advocating for policy change. The goal is informed participation, not withdrawal.
The lesson focuses on active, informed participation — not withdrawal. The four practices described are: verifying before sharing, being intentional about engagement, using reporting tools, and advocating for algorithmic transparency through policy advocacy.
5. The lesson argues that individual responsibility and systemic responsibility for algorithmic harm are:
Correct. The lesson explicitly rejects the either/or framing. Both individual responsibility and systemic responsibility coexist. You are responsible for what you do; the companies that build opaque, addictive systems are responsible for those design choices. Understanding both is necessary.
The lesson argues that both levels of responsibility coexist. Acknowledging systemic responsibility (the company's design choices) doesn't eliminate individual responsibility (what you do with the platform). Understanding both is what allows you to navigate the situation clearly.

Lab 4: Algorithm Designer

You're building a recommendation system. Every design choice you make has trade-offs. Now you have to defend them.

Your Assignment

A major news organization has hired you to design the algorithm for their new news recommendation platform. They want to keep users engaged — that's how they pay the journalists — but they also have a stated mission of an informed public. You have to choose what metric the algorithm optimizes for. Your options include: time-on-site, diversity of perspectives read, accuracy of stories clicked, shares, or some combination.

Your lab partner Amara is a product ethicist who has worked on recommendation systems before. She has seen every rationalization in the book and won't accept vague answers. Every choice you make, she'll ask what you're trading away.

Opening move: Tell Amara which metric you'd optimize for and why. Then tell her what the biggest risk of your choice is — before she points it out herself.
Amara — Product Ethicist
Lab Partner
Before you pitch me your metric — tell me this: the last three recommendation algorithms you've read about in this module all optimized for engagement. Every single one caused harm. Why do you think yours will be different? What's your actual answer to that pattern?

Module 3 Test

Responsibility in a World of Algorithms — 15 questions · Pass at 80%
1. In which country did Facebook's recommendation algorithm play a role in spreading hate speech that contributed to ethnic violence, according to a UN investigation?
Correct. UN investigators found that Facebook had played a "determining role" in spreading hate speech against the Rohingya minority in Myanmar, where the platform was effectively the internet for 18 million users.
The Myanmar case is the documented case. UN investigators found Facebook's algorithm amplified anti-Rohingya content, contributing to ethnic cleansing that displaced over 700,000 people.
2. What was Facebook's recommendation algorithm primarily optimizing for in the mid-2010s?
Correct. Facebook optimized for engagement metrics because engagement kept users on the platform longer, which drove advertising revenue. This made the algorithm indifferent to whether content was accurate or harmful.
Facebook's algorithm was built around engagement: likes, shares, comments, and time on platform. These metrics drove advertising revenue. Accuracy and source quality were not part of the optimization target.
3. A filter bubble forms when:
Correct. A filter bubble is a natural consequence of recommendation algorithms doing their job: they show you more of what you've engaged with, which progressively narrows your information environment to content that confirms what you already believe.
A filter bubble is created by algorithmic personalization, not deliberate censorship or user choice. The algorithm narrows your information environment by optimizing for what you'll engage with — not what's diverse or challenging.
4. ProPublica's 2016 investigation found that the COMPAS algorithm was disproportionately rating Black defendants as high-risk. What was the most likely cause of this disparity?
Correct. COMPAS was not explicitly programmed with racial bias — it learned patterns from historical data that already encoded inequality. More policing in Black neighborhoods meant more arrest records, which the algorithm interpreted as higher risk.
The disparity came from training data bias. Historical criminal justice data reflected decades of unequal policing. The algorithm learned those patterns and reproduced them — without any programmer explicitly building in racial criteria.
5. The "accountability gap" in algorithmic decision-making describes which situation?
Correct. The accountability gap is the situation where harm is caused — but the company, the institution using the algorithm, and the government each deflect responsibility, leaving those harmed with no clear path to redress.
The accountability gap is when harm has clearly occurred from an algorithmic decision, but no single party — the company, the deploying institution, the regulator — takes clear responsibility for addressing it.
6. The EU's General Data Protection Regulation (GDPR) established what right for people affected by algorithmic decisions?
Correct. GDPR's right to explanation means that when an automated system makes a decision affecting you, you are entitled to a meaningful human-readable account of how that decision was reached.
GDPR established the right to explanation — a human-readable account of how an automated decision was reached. This was the first major legal framework to impose accountability obligations on algorithmic decision-making.
7. What is a "deepfake"?
Correct. A deepfake uses deep learning AI to generate or manipulate realistic video, audio, or imagery of real people — creating convincing false records of things they never said or did.
A deepfake is AI-generated media — specifically, realistic video, audio, or imagery that depicts a real person saying or doing something they never actually said or did. The name comes from "deep learning" and "fake."
8. The Gabon case (2019) illustrated which specific risk of the deepfake era?
Correct. The Gabon case showed that you don't need a confirmed deepfake to cause a crisis. The uncertainty alone — could this be fake? — was enough to fuel a coup attempt. Deepfake technology creates a crisis of trust that operates even in its absence.
The Gabon case's key lesson is about uncertainty: the video's authenticity was never confirmed or denied, but the mere possibility that it could be a deepfake contributed to political instability. Doubt itself is a weapon.
9. What does the "liar's dividend" refer to?
Correct. The liar's dividend is the perverse benefit that dishonest people gain from the existence of deepfake technology: they can now deny real, authentic evidence by claiming it was fabricated — and enough people will be uncertain to create doubt.
The liar's dividend is the strategic benefit dishonest actors gain from deepfake technology's existence. They can now claim any authentic video evidence against them is a fake — and given public uncertainty about what's real, this can be a credible defense.
10. Microsoft's Tay chatbot (2016) was shut down after 16 hours because:
Correct. Tay was deliberately targeted by coordinated users who understood it would learn from what they said. It was not programmed to produce hate speech — it learned it from the humans who interacted with it.
Tay was the victim of coordinated abuse: users deliberately fed it racist and hateful content, which it learned and began reproducing. It was obedient to its training data, not malicious — which is precisely what made it vulnerable.
11. When you comment on a misleading post to argue against it, the most likely algorithmic effect is:
Correct. Engagement algorithms track interaction volume, not sentiment. A rebuttal comment is still a comment — still engagement data — which can increase the post's algorithmic visibility. This is one of the counterintuitive dynamics of the engagement optimization model.
Engagement algorithms don't read the content of comments — they count them. A comment arguing against a post still signals engagement, which can increase how widely the algorithm distributes that post.
12. Frances Haugen's 2021 Senate testimony was based on:
Correct. Haugen was a Facebook data scientist who copied thousands of internal research documents before leaving. Those documents showed the company was aware of the harms its algorithm was causing — and the platform kept running anyway.
Haugen's testimony was grounded in internal Facebook documents she had copied. These showed the company's own researchers had documented harm — including the recommendation of increasingly extreme content and the platform's effects on teenage mental health.
13. Training data bias occurs when:
Correct. Training data bias is the mechanism by which AI systems inherit and reproduce existing societal inequalities — not because anyone programmed discrimination explicitly, but because the historical data those inequalities are baked into became the AI's teacher.
Training data bias is when historical inequalities embedded in data get learned and reproduced by an AI. No explicit discrimination needs to be programmed — the AI learns the pattern from data that already encoded unequal treatment.
14. A social media company's own research shows its platform is causing measurable psychological harm to teenage users. The company continues operating the platform without changes. This situation is most analogous to which historical parallel mentioned in the module?
Correct. The module draws this parallel explicitly: a company that knows its product causes harm and continues selling it for profit has a precedent in the tobacco industry's documented concealment of health research.
The module makes the tobacco company parallel explicitly — a company that knows its product causes measurable harm and continues operating it for profit. The ethical question is whether intent matters if the documented harm is the same.
15. This module argues that individual responsibility and systemic responsibility for algorithmic harm are best understood as:
Correct. The module explicitly argues against the either/or framing. Systemic responsibility (how platforms are built and governed) and individual responsibility (how users engage) both exist at the same time. Understanding both is the foundation of genuine digital citizenship.
The module argues these responsibilities coexist. Recognizing that platforms bear systemic responsibility for harmful design doesn't eliminate users' responsibility for how they engage. Both are real, and understanding both is necessary for genuine digital citizenship.