Ray Wang, founder and principal analyst at Constellation Research, believes this tool could be crucial for developers working with large language models (LLMs) to control unwanted responses. The guardrails feature was announced in preview and is expected to be available to all customers sometime next year.
Key takeaways:
- AWS CEO Adam Selipsky announced Guardrails for Amazon Bedrock at AWS re:Invent in Las Vegas, a tool that allows companies to control the language used by large language models.
- Guardrails for Amazon Bedrock allows companies to define topics that are out of bounds for the model, preventing it from answering irrelevant or potentially harmful questions.
- The tool also allows for filtering out specific words and phrases that could be offensive, as well as PII data to keep private data out of the model's answers.
- Ray Wang, founder and principal analyst at Constellation Research, believes this could be a key tool for developers working with large language models to help control unwanted responses.