OpenAI
ChatGPT (GPT-3.5)
Announced Nov 30, 2022
Conversational AI reaches a wider audience
What could it do?
Answer conversational follow ups, explain code and revise a response as the user clarified their request.
What changed?
Human feedback and dialogue training helped turn text generation into an instruction following conversation.
Mainstream adoption
The free research preview offered a direct way to try a conversational assistant.
Where it fell short
It could produce plausible but incorrect answers, respond differently to small prompt changes and be excessively verbose.
Original ChatGPT research preview from November 2022, using a GPT-3.5 family model. Check sources
- Entry type
- Product launch
- Underlying family
- GPT-3.5
- Initial access
- Free research preview
What this meant in practice
The practical change was the feedback loop. You could correct an answer in ordinary language instead of constructing a perfect prompt first. A conversational interface makes iteration easier; it does not make an answer automatically accurate.
Revise a customer reply
Ask for a polite reply, then request a shorter version. Check any promised refund, appointment or policy against your actual records.
Common question
Is this the ChatGPT available today?
No. This page records the original 2022 release, not the current product or its models.
Release facts were checked against the sources below. Performance claims belong to the developers; we have not independently tested these models.
OpenAI: Introducing ChatGPTBenchmark results
No comparable ECI score is available for this release in our source snapshot. Missing scores are never estimated.