Landon Jonquiere Brittan Smith

Title: Landon Jonquiere Brittan Smith and AI Model Retraining.

Introduction

Landon Jonquiere Brittan Smith is an inventor based in San Francisco, California. His patent work concerns methods for improving artificial intelligence language models through feedback on generated responses.

Latest Patents

Smith is associated with 1 patent, titled “Retraining a Large Language Model.” The invention receives an indication that a response from a first AI language model has a problem. The first model generates one or more modified responses, which are provided alongside some or all of the turn-based conversation to a second AI language model. The first model is then retrained based in part on the second model’s output.

Career Highlights

Smith works at Character Technologies Inc. His patent describes a feedback-based approach to retraining language models, using conversational context and evaluations from a second model to help improve responses.

Collaborations

Smith’s listed coworkers include Satyaki Chakraborty and Te-Lin Wu. Their shared professional context at Character Technologies Inc. connects their work to the company’s AI development efforts.

Conclusion

Landon Jonquiere Brittan Smith’s patent focuses on using model-generated feedback and conversation context to retrain an AI language model and improve its responses.