SS
About Me
Frontier AI Paper BriefingsPokebowlClinical Trial EnrollerLittle Human Names
DisclaimersPrivacy PolicyTerms of Use
Privacy Policy·Terms of Use·Disclaimers

© 2026 Silvia Seceleanu

← Back to Explorer
Products·OpenAI·Nov 2022

★16. ChatGPT: Optimizing Language Models for Dialogue

The product that changed everything

Product Announcement
Summary

Launched ChatGPT, a conversational AI fine-tuned from GPT-3.5 using RLHF. It reached 100M users in 2 months — the fastest-growing consumer application in history — and brought AI into mainstream public consciousness.

Key Concepts

GPT-3.5 fine-tuned with RLHF using the InstructGPT pipeline

GPT-3.5 fine-tuned with RLHF using the InstructGPT methodology (SFT → RM → PPO).

The critical innovation was UX, not architecture — a conversational agent anyone could use

The critical innovation wasn't technical — it was the UX decision to present the model as a conversational agent with memory within a session. This made the technology accessible to non-technical users.

1M users in 5 days, 100M in 2 months — fastest-growing consumer app in history

ChatGPT reached 1 million users within 5 days and 100 million within 2 months, surpassing TikTok's growth rate.

Connections

16. ChatGPT: Optimiz…Nov 202212. Evaluating Large…Aug 202113. Training Languag…Mar 202217. Planning for AGI…Feb 202318. GPT-4 Technical …Mar 2023Influenced byInfluences
Influenced by
12. Evaluating Large Language Models Trained on Code (Codex)
Aug 2021
13. Training Language Models to Follow Instructions (InstructGPT)
Mar 2022
Influences
17. Planning for AGI and beyond
Feb 2023
18. GPT-4 Technical Report
Mar 2023