Loading slide

Loading contents...

[██████████░░░░░░░░][████████████████░░░░░░░░░░░░]14 / 25
<back>next

The writer that sees only behind

The second family went elsewhere. A team at OpenAI led by Alec Radford took the writing half on its own, and called the result . It is the family you have actually chatted with.

That split matters more than it sounds. Google had built a translator and kept working on reading. OpenAI kept the half that writes, and bet on making it very large.

Its training game is even simpler, and you have seen it before. Predict the next word. Show it text up to some point, let it guess what comes next, correct it, repeat across an ocean of text.

It only ever sees what came before the word it is guessing, never after, because that mirrors the real task of writing. You produce the next word with only the past to go on.

That one restriction is what makes it a generator. Give it a start and it predicts the next word, then the next, then the next, writing on indefinitely.

Many modern chat systems begin with a causally trained generator, then add later training and application machinery. A later module follows that path.

# citations(1)↓
  1. [1]cdn.openai.com