You’ve used AI for years - spam filters, autocomplete, Netflix picks. So why does ChatGPT feel like a different species?
Look at something you do a hundred times a day: finding the answer to a question.
- Then (rules): you flipped through an encyclopedia or asked around - you did all the digging yourself.
- Learned: Google learned from patterns across billions of pages and ranked the best links for you - but you still had to read them and piece the answer together.
- Today: ChatGPT just gives you the answer, written out, in plain words.
Each step handed more of the thinking to the machine.
Who's doing the thinking?
Step through the eras and watch the work slide from human to machine.
A human wrote every rule by hand. The machine just followed orders - no learning at all.
Step 1 of 4
The three types that matter today
Almost everything called “AI” today is one of three things - and they’re nested, each a subset of the one before.
1. Traditional Machine Learning
- What: algorithms that learn patterns from mostly structured (table-like) data.
- How it learns: a human picks the useful columns - “features” like watch time, genre, time of day - and the model finds the relationship.
- Output: a number or a label - “92% match”, “will watch / won’t”.
- Names you’ll hear: logistic regression, decision trees, random forests, gradient boosting.
- Where you’ve seen it: Netflix’s “Top picks for you” - it learns from the tidy table of what you watched and predicts the next show you’ll binge.
2. Deep Learning
- What: neural networks with many stacked layers. A subset of ML.
- How it learns: feed it raw, messy data (pixels, audio, text) and it discovers the useful features by itself - no manual feature engineering.
- Output: still usually a label/number, but on unstructured data - “there’s a face, here are the eyes”.
- Cost: needs lots of data and serious compute (GPUs).
- Where you’ve seen it: Snapchat’s dog-ear filter - raw camera pixels go in, it finds your face and tracks it as you move, and the ears stay glued on.
3. Generative AI
- What: Deep Learning trained to generate brand-new content, not just label things.
- How it learns: trained on enormous amounts of unlabelled data by predicting missing/next pieces (“self-supervised”).
- Output: new content - text, images, audio, video, code.
- Superpower: one general model handles many tasks it was never explicitly trained for.
- Where you’ve seen it: asking ChatGPT to draft a reply to an awkward text - you give it one line about what you want to say, and it writes the whole message for you, words that never existed before.
Guess the type
Tap the kind of AI you think is behind each everyday moment.
Card 1 of 4
One pile of photos, three jobs
Open your camera roll - one pile of photos, three totally different AI jobs.
Traditional ML predicts which shots you’ll love and auto-favorites them from the metadata.
Deep Learning recognizes faces so you can search “all photos of my dog at the beach.”
Generative AI turns a selfie into a cartoon or paints in a whole new background.
Same photos - predict, recognize, create.
Myth-buster: true or false?
Tap your call on each claim, then see how it lands.
ChatGPT is a totally new kind of tech, unrelated to the AI that came before it.
Netflix recommending shows and ChatGPT writing a reply are doing the same kind of job.
Deep learning figures out the useful clues itself, without a human picking them.
0/3 answered
When to use which
Traditional ML
- In one line
- Learns patterns from tidy tables to predict a number or label.
- Everyday example
- Netflix “Top picks for you”
- Reach for it when…
- Your data is a clean table and you want an explainable prediction.
Deep Learning
- In one line
- Finds its own features in raw images, audio, or text.
- Everyday example
- Snapchat face filters
- Reach for it when…
- The input is raw pixels, audio, or free text - not a table.
Generative AI
- In one line
- Creates brand-new content from a prompt.
- Everyday example
- ChatGPT drafts your reply
- Reach for it when…
- You need to make or transform content - text, images, code.
Pick the right tool
Drag each task onto the approach that fits best.
Traditional ML
Predict from tidy tables
Deep Learning
Perceive raw images / audio
Generative AI
Create & summarize language
0/6 correct
- AI evolved like finding an answer: encyclopedia/asking around -> Google ranking links -> ChatGPT just answering, each handing more thinking to the machine.
- They're nested: GenAI is deep learning; deep learning is ML; ML is AI.
- Match the tool to the job: Netflix-style tables -> ML; raw pixels/audio like face filters -> DL; creating content like ChatGPT writing your reply -> GenAI.