In a recent announcement, Google CEO Sundar Pichai revealed plans to launch two “big language models” in the near future. LaMDA (Language Model for Dialogue Applications) and PaLM (Predicting an Agent’s Long-term Memories) are both set to be released in the next few weeks and months. The company is racing against ChatGPT, another generative AI technology that could pose a threat to Google’s dominance in the AI space. Let’s take a look at what these models have to offer.
LaMDA is essentially a conversation engine that has been trained on millions of conversations taken from books, films, and other sources. According to Pichai, it can generate more “factual” answers than its competitors by leveraging complex natural language processing techniques. The goal of LaMDA is to make interacting with computers easier by providing conversational responses instead of just listing search results. It will be integrated into Google Search first, but the company also plans to use it across many of their products such as Gmail and Docs in the future.
PaLM is a more ambitious project that seeks to emulate human memories by predicting what an agent would remember about past events in order to better understand current conversations. PaLM isn’t ready for release yet since it’s still very much in its early stages, but Pichai hinted at some potential applications such as allowing robots to become aware of their environment over time or helping voice assistants better understand spoken commands.
Google’s upcoming language models are set to revolutionize the way we interact with machines, making them more accessible and responsive than ever before. LaMDA will soon be available for public use while PaLM continues development over the coming months—so stay tuned! With these technologies under its belt, Google looks poised to remain at the forefront of generative AI for years to come.
