Artificial Intelligence (AI)

OpenAI just announced that ChatGPT now has memory!

🇧🇷🇺🇸-  Book: Making Things Think: https://holloway.com/mtt. Investor in Wander, Carry, Footprint, Merkle Manufactory (Farcaster), Dynamic, Paragraph.

Is it really "memory" or that they are managing data and resubmitting data to the LLM based upon cosine similarity of the prompt - i.e. they baked a vector database into ChatGPT

Polyglot developer (Rust, Python, TypeScript these days) | 075 Member | AI Dev | Godot and Unity dev | Founder of Block Imperium Games /blockimperium

AI Desk Research Tool/Agent @deskrex , http://lp.deskrex.ai , Onchain-referral on web3 social @Refidentity , https://refid.xyz

Depends on if you rely on embedding. If not, any in-memory cache would be sufficient.

Do you know any way to create the real "memory"  not using the vector db and RAG solution?