LaMDA

LaMDA, Language Model for Dialogue Applications, is the family of conversational large language models Google unveiled at I/O 2021. It grew out of Meena, a 2.6B-parameter chatbot Google showed in January 2020, scaled up with more data and compute. LaMDA was trained specifically for open-ended dialogue, with Google emphasizing sensibleness, specificity, interestingness, plus safety and groundedness: the ability to cite external sources. The I/O demo had LaMDA role-playing as Pluto and a paper airplane.

LaMDA became globally famous in June 2022 when Google engineer Blake Lemoine claimed it was sentient, a claim Google and the research community rejected while acknowledging it raised useful questions about the Turing test. In February 2023 Google launched Bard on a lightweight LaMDA version to counter ChatGPT, and by late 2023 the LaMDA branding faded as Brain and DeepMind merged and the Gemini family took over. LaMDA was always proprietary: no weights were released, so openWeights is false. Its dialogue-focused training techniques underpin Google's later conversational systems.