Shulou(Shulou.com)11/24 Report--

Thanks to CTOnews.com netizens Wu Yanzu in South China for the delivery of clues! CTOnews.com, April 26, the development of artificial intelligence (AI) is changing with each passing day, but there are also some difficulties, such as AI models sometimes have "hallucinations", that is, telling the wrong facts, involving harmful topics, or causing security risks. To solve this problem, Nvidia released a new software called NeMo Guardrails on Tuesday, which can help software developers set "guardrails" on AI models to prevent them from producing bad output.

NeMo Guardrails is a software layer between the user and the AI model that can be intercepted and modified before the model outputs bad content. For example, if developers want to create a customer service chat robot, they can use NeMo Guardrails to restrict it to talking about related products, not competitors' products or other irrelevant topics. If the user asks such a question, the robot can lead the conversation back to the topic the developer wants.

Nvidia also offers another example, a chat robot used to answer internal corporate human resources questions. In this example, Nvidia can use NeMo Guardrails to prevent ChatGPT-based robots from answering questions about the company's finances or accessing the private data of other employees.

In addition, the software can use one AI model to detect the "hallucinations" of another AI model and verify the answer of the first AI model by asking questions about the second AI model. If the two models give inconsistent answers, the software will return an "I don't know" answer.

Nvidia also said the "guardrail" software also helps improve security by forcing the AI model to interact only with third-party software on the whitelist.

NeMo Guardrails is open source and available through Nvidia services for commercial applications, and developers will use the Golang programming language to write custom rules for the AI model.

According to CTOnews.com search, other AI companies, including Google and Microsoft-backed OpenAI, also use a method called reinforcement learning to prevent harmful output from LLM applications. This approach uses human testers to create data about which answers are acceptable or unacceptable, and then uses that data to train the AI model.

Nvidia is turning its attention to AI, which now dominates the market for chips needed to create the technology, and its shares have risen 85 per cent so far in 2023, making it the biggest gainer in the s & p 500s.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.