Class 7 · CBSE AI · Strand A — Systems Thinking
Algorithmic cascades — when AI errors amplify each other
When one AI's output feeds another, small errors can cascade — like the 2010 Flash Crash. For Class 7.
What this concept actually says
- An algorithmic cascade occurs when the output of one AI system becomes the input of another, amplifying errors across the chain
- Cascades are especially dangerous when systems operate at speed and scale without human checkpoints between them
- Designing circuit breakers — automatic pauses that trigger human review — is a key safeguard against cascades
An analogy your child will recognise
Train signalling cascade
Imagine a signal failure at one station that automatically re-routes trains. The re-routing creates congestion at the next station, which triggers another automatic re-route, which now blocks a third station. Within an hour, a single signal error has cascaded into a jam across three districts — each automated response made perfect sense locally but worsened the system globally.
Rumour in a mela
A shopkeeper at a mela hears a rumour that tomorrow's mela is cancelled. He tells three vendors. Each tells five more. By evening, 200 vendors have packed up and left — and nobody checked the original source. Each person acted rationally on the information they received. The cascade was perfect; the original input was false.
Common misconceptions to watch for
- Cascades only occur when an AI makes a mistake — a cascade can also be triggered by a correct AI decision that has unexpected downstream effects.
- Adding more AI systems to check each other's work prevents cascades — in practice, it can create more cascade pathways.
Key facts in one breath
- The 2010 Flash Crash erased nearly $1 trillion in market value in minutes before partially recovering — triggered by algorithmic trading cascades.
- A cascade occurs when AI system outputs feed directly into other AI systems without human validation between them.
- Speed and scale amplify cascades: what takes humans hours to notice, interconnected AIs can propagate in seconds.
- Circuit breakers are automatic pause mechanisms that halt AI action when certain thresholds are exceeded and require human review before resuming.
How Dhee Learning teaches this — the 3-stage question loop
Every Dhee Learning session for this concept follows three stages. We share the questions Dhee actually asks, so you can hear what a session sounds like.
Stage 1 — Surface
A news app runs three AIs in a chain — translate, then moderate, then recommend — with no human between. If AI 1 mistranslates one word, why is that dangerous for the whole app?
Rote answer
"The AIs are connected."
Understood
"AI 1's output is AI 2's input, and AI 2's is AI 3's — a cascade. AI 2 never saw the original, so it can't tell the word is wrong and passes it on; AI 3 recommends it. With no human between and at machine speed, one small error reaches a million screens before anyone can catch it."
Stage 2 — Reasoning
In the news app's translate to moderate to recommend chain, explain how a single mistranslation at stage 1 travels all the way to a million readers — and where a human break would stop it earliest.
Follow-up Dhee may use: Why is a smarter single AI not the fix here, when each AI in the chain is already working correctly?
Stage 3 — Application
A hiring app chains three AIs with no human between: a resume-screener, then an interview-scheduler, then a ranking AI. Design one circuit breaker for this chain — say exactly where it goes, what triggers it, and what the human does.
Misconception Dhee watches for: Child places the circuit breaker at the END of the chain (after the harm is done) or thinks a bigger AI replaces the human break, rather than placing an early automatic pause between the AIs.
Related concepts
Want your child to actually understand this?
Dhee turns this concept into a short spoken lesson — teaching, listening, and probing — so your child builds the idea themselves.
Frequently asked questions
What is when ai feeds ai — explained for kids? +
When one AI's output feeds another, small errors can cascade — like the 2010 Flash Crash. For Class 7.
What's the most common mistake children make about this concept? +
Cascades only occur when an AI makes a mistake — a cascade can also be triggered by a correct AI decision that has unexpected downstream effects.
How does Dhee Learning teach this in a Class 7 session? +
Dhee opens with a question — for example: "A news app runs three AIs in a chain — translate, then moderate, then recommend — with no human between. If AI 1 mistranslates one word, why is that dangerous for the whole app?" — listens to your child's answer, then probes the reasoning behind it. The session ends when the child can apply the idea to a brand-new situation, not just recall it.