Google has focussed on five areas that it continues to work on: accuracy, bias, persona, false positives and false negatives, and refining Bard model.
Since Bard is in its early stages, it can generate responses that contain inaccurate or misleading information at times, Google said.
Google explained that Bard is based on a technology that predicts the next word or sequences of words. So there may be times that the model may return inaccurate information.
All large language models (LLMs) use training data, including from publicly available sources, to generate answers.
Since the data is taken from multiple sources, it brings diversity of perspectives and opinions. Therefore, for subjective topics, such as politics, Google said that Bard is designed to provide users with multiple perspectives.
Google has also highlighted that Bard might, at times, generate responses that seem to suggest it has opinions or emotions. This is because Google trained the model in a language that people use to reflect the human experience.
Google said it has limited Bard from generating problematic responses (such as harmful/offensive content) to prompts it hasn’t been trained to.
Google noted that users must be 18 years old or older to try because Bard, despite having technical guardrails, may sometimes generate problematic information.
Google said that it wants users to test the limits of what Bard can do and attempt to break Bard's protections so that it can be trained better.
Thanks For Reading!