Developers: | |
Date of the premiere of the system: | May 2021 |
Technology: | Speech technologies |
2021: The announcement of an AI system that learns during a conversation and begins to conduct a conversation indistinguishable from a person
At the end of May 2021, Google announced a new AI system, which is being trained during the conversation and begins to conduct a conversation indistinguishable from a person. The new model is called LaMDA (Language Model for Dialogue Applications, "language model for dialog applications"). Google called the new system "revolutionary communication technology," which will allow users to dialogue more naturally and openly with AI assistants.
Language models help computers process and reproduce natural language, that is, the language of people. For example, thanks to such models, an auto-replacement system works. Usually, language models are trained only on texts obtained from the Internet. LaMDA is trained in dialogue, that is, knows how to capture nuances unique to open conversations.
In his preview, Google showed a video of two conversations with the model, where LaMDA answered questions, "pretending" to be the dwarf planet Pluto, as well as a paper airplane. The model was able to quote real events and facts on two topics, such as the visit of the New Horizons probe to Pluto in 2015. When asked what the dwarf planet would like to tell people about itself, Pluto replied: "I want people to know - I am not just an accidental ice ball. In fact, I am a beautiful planet. "
LaMDA "is designed to discuss any topics," said Google CEO Sundar Pichai, explaining that the model is still under development but is already being used internally. |
Google studies the "interest" of the model (assessing the insight, surprise and/or wistfulness of the answers), and also works to ensure that the LaMDA meets the standards of fairness, accuracy, security and privacy. Pichai did not provide details about the data used for training, prohibited topics or specific precautions applied by the company.[1]