Lies at scale: in brief
Lesson 9 is the last lesson, and it closes the whole track. Lesson 8 asked what AI does to your work; this one asks what AI does to something even more personal, what you are able to trust. It adapts the mis and disinformation session of the Harvard Kennedy School course this track is built on, created by Sharad Goel, Dan Levy, and Teddy Svoronos, and it takes a frightening fact (the same tool that drafts your emails can pour convincing false words, images, and voices into the world faster than anyone can check them) and turns it, by the end, into a reason to feel steadier. The point of this track was never fear. It was standing on solid ground.
The capability: after this lesson you can tell a mistake from a lie, recognize a likely hallucination and check it before it costs you, feel the tug of a fake built to move you and slow down instead of sharing, spot the liar’s dividend when someone waves real evidence away, and use lateral reading to decide whether a claim, image, or recording deserves your trust. All of it is recognition and resilience: you are learning to check and to keep your footing, never to produce anything false.
What the lesson covers. First, two plain terms and the line between them: misinformation is false information that spreads without anyone meaning to deceive, disinformation is false information someone spreads on purpose, and the whole difference is intent. Then the machine’s own false claims: recalling from Lesson 1 that a generative model writes one word at a time by predicting what is likely to come next, the lesson gives that failure its name, a hallucination, a confident, fluent, false claim, and hands you the two questions to meet it with (does it matter whether this is true, and can I check it). Then disinformation on purpose: how cheap fakes got, the three-step path a lie must travel (someone makes it, someone spreads it, someone believes it) used as a recognition frame for where the chain breaks, why distribution and not creation is still the bottleneck, and the personalized deepfake aimed at a face you trust, answered with the same move you learned for the voice-clone scam in Lesson 6: verify through a separate channel you already trust. Then the deepest harm, the liar’s dividend, where once anything can be faked, real evidence gets waved away as fake, already reaching the courts, answered always with verify, do not dismiss. It closes on the four-actor resilience toolkit (the companies that build the tools, the platforms that carry the content, governments, and you), where every fix higher up the chain leaks, which is exactly why your own habit, lateral reading, is the part that holds.
Why this order, and where it leaves you. The lesson moves from the smallest, most private worry (is this one thing I am looking at true?) outward to the civil-society response, and lands on you as the reliable one rather than the last resort. Because this is the final lesson, it does not point forward. It looks back across all nine lessons and hands you what you carry out: not a shield that blocks every fake, because no one has that, but your own judgment, sharpened. The practice is a recognition workout, not a production exercise: you sort scenarios by what kind of falsehood each is, drill lateral reading on a benign claim, and, in Clawless, hold a plain conversation with a model and fact-check its answer against a trusted independent source, treating the model as a subject to verify and never an oracle. Lesson 1 said that once you understand the trick, the fear starts turning into fluency. This lesson is where that lands.