@simonw

–

18 September 2025 21h33

Yeah, I'd argue that ChatGPT itself has been an "agent" ever since they added Code Interpreter to it back in early 2023

OpenAI’s Inconsistent Agent Definitions Need Clarification

By

–

18 September 2025 21h31

The worst offenders for constantly redefining agents with new, vague and inconsistent definitions are OpenAI themselves https://
simonwillison.net/2025/Sep/18/ag
ents/#openai-need-to-get-their-story-straight
…

Defining AI Agents: Why Vague Definitions Fall Short

By

–

18 September 2025 21h26

That's one of the definitions I don't like, I think it's too vague to be useful A thermostat is an "AI agent" at that point

LLM Library Gets Tool Calling for Agentic Framework

By

–

18 September 2025 21h25

I added tool calling to my LLM CLI/Python library a few months ago which upgraded it to an "agentic framework", but I'm only now ready to admit that's what it is because I didn't like using the term agents!

AI Agents Versus Human Agency: Definitional Critique

By

–

18 September 2025 21h20

Includes this note about why agents as human replacements is my least favorite definition – because unlike AI agents, humans have agency!

Agency: The Critical Difference Between AI Agents and Humans

By

–

18 September 2025 21h17

In my post I argue that "agency" is the one thing that AI agents don't have and humans do! https://
simonwillison.net/2025/Sep/18/ag
ents/#agents-as-human-replacements-is-my-least-favorite-definition
…

Defining LLM Agents: Running Tools in Loops for Goals

By

–

18 September 2025 21h14

I'm ready to accept a definition of "agent" that I think is widely-enough agreed upon to be useful: An LLM agent runs tools in a loop to achieve a goal This is a big piece of personal character development for me! I've been dismissing the term as hopelessly ambiguous for years