# From training data to your feed: the AI pipeline

> A process map exported from forks.world (https://forks.world). Author: Brian. Source: https://forks.world/p/cmv0kq8i1003qjv06h5p25k0q
> It describes who does what, in what order, and with which technology. Read the plain-language sections first; the JSON at the end is the exact, complete data.

## Summary

- 8 steps across 1 layer
- Persona: AI labs, Platforms, You

## How to read this file

- A process is made of **layers**. The main layer is the top level; any step of kind `subprocess` opens a detail layer with its own lanes and steps.
- **Lanes** are either persona (people/roles) or technology (systems). Each step belongs to one lane; `using` lists tools a person uses for that step.
- Step kinds: `start`, `task`, `decision` (its arrows carry the answer as a label), `subprocess`, `end`.

## From training data to your feed: the AI pipeline

Lanes:
- AI labs: person / role (id `lab`)
- Platforms: person / role (id `platform`)
- You: person / role (id `you`)

Steps, in the order they run:

1. **Train a model** [subprocess; AI labs; id `a1`], links to the published process "How a chatbot like ChatGPT or Claude is made" (https://forks.world/p/cmv0kppjs003jjn06e9er1os6; full text at https://forks.world/api/public/cmv0kppjs003jjn06e9er1os6/export)
   - then It writes text
2. **It writes text** [subprocess; AI labs; id `a2`], links to the published process "How a language model writes one word at a time" (https://forks.world/p/cmv0kpq1p003ojn06v04h6iiq; full text at https://forks.world/api/public/cmv0kpq1p003ojn06v04h6iiq/export)
   - then It makes images
3. **It makes images** [subprocess; AI labs; id `a3`], links to the published process "How AI turns a sentence into an image" (https://forks.world/p/cmv0kpqkw003tjn064o6q1msp; full text at https://forks.world/api/public/cmv0kpqkw003tjn064o6q1msp/export)
   - then The feed ranks what you see
4. **The feed ranks what you see** [subprocess; Platforms; id `a4`], links to the published process "How a social feed decides what you see" (https://forks.world/p/cmv0kprm90043jn06dfxesqhy; full text at https://forks.world/api/public/cmv0kprm90043jn06dfxesqhy/export)
   - then Is it real?
5. **Is it real?** [decision; You; id `a5`]
   - if "Probably" go to Share it; if "Maybe not" go to Check the source
6. **Share it** [end; You; id `a6`]
7. **Check the source** [subprocess; You; id `a7`], links to the published process "How a deepfake is made and caught" (https://forks.world/p/cmv0kpr3z003yjn06ou1iinlz; full text at https://forks.world/api/public/cmv0kpr3z003yjn06ou1iinlz/export)
   - then Share or skip
8. **Share or skip** [end; You; id `a8`]

```mermaid
flowchart LR
  subgraph l_lab["AI labs"]
    n_a1[["Train a model"]]
    n_a2[["It writes text"]]
    n_a3[["It makes images"]]
  end
  subgraph l_platform["Platforms"]
    n_a4[["The feed ranks what you see"]]
  end
  subgraph l_you["You"]
    n_a5{"Is it real?"}
    n_a6(["Share it"])
    n_a7[["Check the source"]]
    n_a8(["Share or skip"])
  end
  n_a1 --> n_a2
  n_a2 --> n_a3
  n_a3 --> n_a4
  n_a4 --> n_a5
  n_a5 -->|"Probably"| n_a6
  n_a5 -->|"Maybe not"| n_a7
  n_a7 --> n_a8
```

## Full data (JSON)

```json
{
  "title": "From training data to your feed: the AI pipeline",
  "maps": {
    "m_root": {
      "id": "m_root",
      "lanes": [
        {
          "id": "lab",
          "name": "AI labs",
          "type": "person"
        },
        {
          "id": "platform",
          "name": "Platforms",
          "type": "person"
        },
        {
          "id": "you",
          "name": "You",
          "type": "person"
        }
      ],
      "steps": [
        {
          "id": "a1",
          "kind": "subprocess",
          "lane": "lab",
          "link": {
            "id": "cmv0kppjs003jjn06e9er1os6",
            "scope": "public",
            "title": "How a chatbot like ChatGPT or Claude is made"
          },
          "next": [
            {
              "to": "a2"
            }
          ],
          "uses": [],
          "label": "Train a model"
        },
        {
          "id": "a2",
          "kind": "subprocess",
          "lane": "lab",
          "link": {
            "id": "cmv0kpq1p003ojn06v04h6iiq",
            "scope": "public",
            "title": "How a language model writes one word at a time"
          },
          "next": [
            {
              "to": "a3"
            }
          ],
          "uses": [],
          "label": "It writes text"
        },
        {
          "id": "a3",
          "kind": "subprocess",
          "lane": "lab",
          "link": {
            "id": "cmv0kpqkw003tjn064o6q1msp",
            "scope": "public",
            "title": "How AI turns a sentence into an image"
          },
          "next": [
            {
              "to": "a4"
            }
          ],
          "uses": [],
          "label": "It makes images"
        },
        {
          "id": "a4",
          "kind": "subprocess",
          "lane": "platform",
          "link": {
            "id": "cmv0kprm90043jn06dfxesqhy",
            "scope": "public",
            "title": "How a social feed decides what you see"
          },
          "next": [
            {
              "to": "a5"
            }
          ],
          "uses": [],
          "label": "The feed ranks what you see"
        },
        {
          "id": "a5",
          "kind": "decision",
          "lane": "you",
          "next": [
            {
              "to": "a6",
              "label": "Probably"
            },
            {
              "to": "a7",
              "label": "Maybe not"
            }
          ],
          "uses": [],
          "label": "Is it real?"
        },
        {
          "id": "a6",
          "kind": "end",
          "lane": "you",
          "next": [],
          "uses": [],
          "label": "Share it"
        },
        {
          "id": "a7",
          "kind": "subprocess",
          "lane": "you",
          "link": {
            "id": "cmv0kpr3z003yjn06ou1iinlz",
            "scope": "public",
            "title": "How a deepfake is made and caught"
          },
          "next": [
            {
              "to": "a8"
            }
          ],
          "uses": [],
          "label": "Check the source"
        },
        {
          "id": "a8",
          "kind": "end",
          "lane": "you",
          "next": [],
          "uses": [],
          "label": "Share or skip"
        }
      ],
      "title": "From training data to your feed: the AI pipeline",
      "parent": null
    }
  }
}
```