Learn

Learn AI, one idea at a time

AI is more than a chatbot and an image maker. Short lessons on the ideas behind the news, in simple words, each with why it matters. They grow from the stories in our daily edition and in Le Café Noir.

Agents

• What is an AI agent?An AI that takes actions toward a goal, not only answers.
• What is Agentic Tasks?Asking a chatbot a question is like asking a friend for a recipe. An agentic task is like asking that friend to go shopping, cook the whole meal, and clean up, all alone while you are out.
• What is Compaction?Imagine your notebook is full, so you write a short summary of the old pages on one new page and throw the old pages away. Now you have room again, but some little details are gone.
• What is Pull Request?A pull request is like handing your draft to an editor before it goes in the newspaper. The editor checks it, asks for fixes, and only then publishes it.

Security

• What is prompt injection?Hidden text that tricks an AI into following an attacker.
• What is a CVE?A public ID for one known software flaw, and why the count matters.
• What is Black October?A record month of software flaws, and why fixing is the gap.
• What is Coordinated Disclosure?Imagine you find a hole in your neighbour's fence. You tell them quietly first, so they can fix it before you tell everybody, and before anyone sneaks through.
• What is Exploit?A bug is a broken lock on a door. An exploit is someone actually opening the door with a trick and walking in.
• What is Jailbreak?A jailbreak is like a kid finding the exact words that trick the babysitter into allowing candy before dinner, even though the parents said no.
• What is Remote Code Execution?Imagine a stranger who can shout orders through your window, and your house robot obeys them like they came from you. Remote code execution is when a stranger far away can make someone else's computer follow their orders.
• What is Reverse Engineering?It is like taking apart a toy car with a screwdriver to see how the wheels turn, because nobody gave you the instructions.
• What is Sandbox?A sandbox is like a playpen for a puppy: it can run, chew and explore inside, but the fence keeps it away from the rest of the house.

Models and how they work

• What is Benchmark?A benchmark is like a taste test where every cook makes the same dish, so a judge can tell who did best. AI models get scored this way too.
• What is Frontier Model?A frontier model is like the newest, fastest race car that only a few big teams can build. It can do amazing things, but it can also crash harder than a normal car.
• What is Mixture of Experts?Imagine a big hospital with many specialists, but for each question you only see the two doctors who know that subject best. The whole hospital is big, but each answer only needs a few of them.
• What is Open-weight model?A chatbot like ChatGPT is like eating at a restaurant: you get the meal, but the kitchen stays theirs. An open-weight model is like getting the finished cake to take home, so you can keep it, slice it, and even add your own icing.
• What is Prefilling?It is like starting a sentence for your friend, "Once upon a time, the dragon...", and then letting them finish the story. The AI keeps going from the words you put in its mouth.
• What is Training data?Training data is like all the books a child reads before they can write their own stories. The AI reads a giant library first, then it writes by copying the patterns it saw.
• What is Weights?Weights are like the knobs on a giant music mixer. During training, the AI turns millions of little knobs until the music sounds right, and those knob positions are what it "knows."
• What is World model?A world model is like a child playing with toy cars who already knows that if you push a car off the table, it falls. It imagines what happens next before it happens.

Data and privacy

• What is KV Cache?When you tell a long story, you keep the earlier parts in your head so you don't start again from the beginning each time. The KV cache is the AI's little notepad for remembering what it already read and wrote in your conversation.

Risks and safety

• What is Red-teaming?Red-teaming is like asking your big brother to try to sneak into your pillow fort, so you can see where the walls are weak before anyone else finds them.
• What is Voice cloning?Voice cloning is like a parrot that hears you talk for a little while and then can say anything in your voice, even things you never said.

Get The AI Daily Grind by email, every morning, in simple words.