Module 1 · AI foundations · ~59 min · Engine AI-v6

Teach a machine
to learn.

Every model family that matters, opened up and run with real telecom numbers. You will compute a regression by hand, watch gradient descent roll downhill, read a confusion matrix like a ledger, follow backpropagation’s wave of blame, meet attention, LLMs and reinforcement learning — and learn the discipline (splits, leakage, drift) that separates a demo from a deployment.

▶  Play the module ✦  Open the playground
29chapters · ~59 min
15model families
35scenes · 140+ beats
2live playgrounds
Watch · Module 1 · ~59 min

One hour, every model family — run on live telecom numbers

The stage renders every idea as it is spoken: the loss landscape with its rolling ball, the forest voting, k-means centroids gliding, the forward pass flowing and the backpropagation blame-wave glowing red, attention threading its lines, and the Q-learning update ticking through real arithmetic. English karaoke subtitles, word by word.

output · live render — Module 1 · The Learning Machinecue-synced engine
CT-AITEL-1000 · Module 1 · AI in Telecom Optimization
The Learning Machine — what AI actually is
0:00 / 0:00
Playground · touch the math

Run gradient descent yourself — then fire a neuron

The same numbers the video teaches. Drag the learning rate and watch descent converge — or push it past the stability limit and watch the loss explode. Then drive a single neuron’s three inputs and read tomorrow’s degradation risk off the sigmoid.

⭐ gradient-descent playgroundlive math
2.0
w = 0.00 · loss = 21,859 — the four measured points: 20%→206 · 40%→160 · 60%→118 · 80%→74 Mbps (b fixed at 249). Stability limit here is η ≈ 3.3×10⁻⁴ — drag past it and watch the loss explode.
⭐ fire a neuron — tomorrow’s degradation riskσ(w·x+b)
+1.5 +1.2 −1.0
Reference · the model families

Fifteen families, one card each — or open the Model Analyzer ↗

What it predicts, the loss it descends, where it shines on network data, and the trap that bites in production. Filter as you revise.

Reference · the vocabulary

Every term of Module 1, in one searchable table

From loss functions to leakage — the working vocabulary of applied machine learning, each with its formula or origin and its one-line meaning.

termfamilyformula / originwhat it means
Mastery check

The Module 1 mastery check

Ten questions drawn from a pool of nineteen, reshuffled on every load — answers and order both. Score 70% or better before moving to Module 2; the network is waiting.

Q 1 / 10
0 ✓