Generative artificial intelligence startup AI21 Labs Ltd. said today it has created a new plug-and-play tool that will forever banish the plague of AI hallucinations by ensuring chatbot responses are based on an organization’s own knowledge base.

The new tool is called Contextual Answers, and it can be embedded into digital assets to enhance the efficiency and accuracy of information queries with generative AI, the startup said.

AI hallucinations occur when generative AI-powered chatbots simply fabricate their answers because they cannot find the information required to respond to a prompt truthfully. They can do a lot of damage, with one recent example involving a law professor who was falsely accused by ChatGPT of sexually harassing one of his students. ChatGPT cited a 2018 article in The Washington Post as the source of this claim — but the problem was that no such article existed, and the professor in question had never faced such accusations.

While that must have been extremely frustrating for the individual concerned, AI hallucinations can cause other problems. For example, if a company uses generative AI to power customer service chatbots, they want to ensure that everything it says is correct, and doesn’t provide wrong information to consumers.

With Contextual Answers, AI21 Labs says it’s able to increase the reliability of generative AI because it has been engineered to return only completely accurate and relevant responses to queries based on source documents uploaded by the organization.

Organizations can upload unlimited amounts of any data they see fit, such as an internal knowledge base, the content from a help center library, financial reports or technical manuals, and their chatbots will only use this information as a source for their responses. The company said documents can be uploaded by anyone, even those without prior AI experience. It said its model is capable enough to understand misspelled and missing words, enabling it to respond to queries that contain grammatical errors.

Contextual Answers supports multiple use cases, the company said. For instance, it can be used to power customer service bots that respond to customers using internal organization documentation to ensure factual accuracy. It can also be used by financial services providers, helping to analyze business documents such as financial reports and call logs. Other use cases include education, where it can help teachers and students to quickly research and find answers in large databases, and legal services and insurance, helping users to quickly surface relevant legal documents such as legislation, regulations and compliance guidelines.

AI21 Labs said Contextual Answers will be integrated into its flagship AI tool Wordtune, which is an AI-powered writing companion that can understand content and meaning to help people write down their ideas in a clearer and more compelling way.

Co-founder and co-Chief Executive Ori Goshen said initial fears around AI are beginning to wane as people come to realize its benefits, but companies still need to prevent AI hallucinations. “We expect companies will get smarter about how best to implement AI in their respective workflows and embrace the right tools,” he said. “That’s why we’re so excited about the launch of Contextual Answers, further positioning us to serve as a key piece of any company’s AI strategy.”

Photo: Xu Haiwei/Unsplash