Please stop forcing Clippy on those who want Anton
Update: Steven Sinofsky agrees: 1) Schticks get tired, 2) Helpfulness gets less helpful as skill level improves. Society advances by the number of operations you can do without thinking, not by the number of times you are complimented on your brilliance. The original title of this post was “Please don’t put memory everywhere”. The proximate cause was the rollout of improved memory in ChatGPT1 this month. Many many many people at OpenAI hyped it up, the priesthood anointed it “possibly life-changing”, and Rich Sutton called it the Era of Experience2. However, the number of serious people reporting issues (bad assumptions, hesitation to use, social calculation) have caused even some of the OpenAI folks to concede3 that we may need an “innie” and “outie” ChatGPT with severed memories between them. But this goes deeper than memory. Of course there’s a relevant Silicon Valley clip: To quote: Gilfoyle: It's bad enough it has to talk, does it need fake vocal tics like “uh”? Dinesh: Well that
Update: Steven Sinofsky agrees: 1) Schticks get tired, 2) Helpfulness gets less helpful as skill level improves. Society advances by the number of operations you can do without thinking, not by the number of times you are complimented on your brilliance. The original title of this post was “Please don’t put memory everywhere”. The proximate cause was the rollout of improved memory in ChatGPT1 this month. Many many many people at OpenAI hyped it up, the priesthood anointed it “possibly life-changing”, and Rich Sutton called it the Era of Experience2. However, the number of serious people…
related reading
- Child’s Play, by Sam Krissharpers.org
- AI Friends Too Cheap To Meterjasmi.news
- AI #155: Welcome to Recursive Self-Improvementthezvi.substack.com
- Gossip is Goodfuturelovers.com
- Clippy: A Little History of Little Guys - thejaymo.netthejaymo.net
- The World Still Hasn’t Made Sense of ChatGPT - The Atlantictheatlantic.com
- the post-literate societyjasmi.news
- Doomprompting Is the New Doomscrolling.workingtheorys.com
- Why Chatbots Are Not the Future of Interfaceswattenberger.com
- Expanding on what we missed with sycophancy | OpenAIopenai.com
- AI #87: Staying in Character - by Zvi Mowshowitzthezvi.substack.com
- Does your AI perform badly because you — you, specifically — are a bad person?nataliercargill.substack.com