Would I Use LLMs to Rebuild Twitter's Dynamic Product Ads? Yes and No!

This story was originally published on HackerNoon at: https://hackernoon.com/would-i-use-llms-to-rebuild-twitters-dynamic-product-ads-yes-and-no.
How LLMs improve product recommendations at scale - and where classic ML still wins. Lessons from building Twitter's ad system on what actually matters.
Check more stories related to machine-learning at: https://hackernoon.com/c/machine-learning. You can also check exclusive content about #ai, #llms, #twitter, #ecommerce, #ads, #embedding-models, #dynamic-product-ads, #hackernoon-top-story, and more.

This story was written by: @manoja. Learn more about this writer by checking @manoja's about page, and for more stories, please visit hackernoon.com.

Twitter’s Dynamic Product Ads system was built to match users to products they are most likely to buy. The system used product and user embeddings with classic ML models to serve personalized ads at Twitter’S scale. The approach was very textbook (for 2022 at least). Would this approach have been different in 2026 with AI and LLMs?