AI21 Labs introduces anti-hallucination feature for GPT chatbots.

AI21 Labs introduces anti-hallucination feature for GPT chatbots.

Blockchain Industry: Addressing Trustworthiness and Mitigating Hallucination in AI Models

The blockchain industry has been witnessing paradigm-shifting advancements in recent years, with the introduction of technologies like artificial intelligence (AI) and large language models (LLMs). One of the latest developments in this field is the launch of “Contextual Answers” by AI21 Labs, a question-answering engine specifically designed for LLMs.

AI21 Labs’ Contextual Answers engine allows users to upload their own data libraries, enabling them to restrict the outputs of LLMs to specific information. This addresses a significant concern in the AI industry – the lack of trustworthiness in AI models. While products like OpenAI’s ChatGPT have been groundbreaking, businesses often hesitate to adopt them due to the potential for incorrect or “hallucinated” responses.

Research indicates that employees spend nearly half of their workdays searching for information. This presents a massive opportunity for chatbots capable of performing search functions. However, most chatbots available today are not tailored for enterprise use. Recognizing this gap, AI21 developed Contextual Answers to bridge the divide between general-purpose chatbots and enterprise-level question-answering services.

According to AI21, Contextual Answers allows users to steer AI answers without the need for retraining models. This mitigates some of the major obstacles to AI adoption, such as cost, complexity, and the lack of specialization of models in organizational data. By enabling users to pipeline their own data and document libraries, Contextual Answers empowers businesses to leverage AI technology effectively.

One of the challenges associated with the development of useful LLMs, such as OpenAI’s ChatGPT or Google’s Bard, is teaching them to express a lack of confidence. In typical scenarios, when a user queries a chatbot, it tends to provide a response even if it lacks sufficient information in its dataset to give accurate information. Rather than responding with a low-confidence answer like “I don’t know,” LLMs often generate information without any factual basis, leading to what researchers call “hallucinations.”

AI21’s Contextual Answers aims to solve the hallucination problem entirely. It achieves this by either providing information only when it is relevant to user-provided documentation or by not outputting anything at all. This approach ensures that the AI model does not generate false or misleading information, thereby enhancing its trustworthiness.

The finance and legal sectors, where accuracy is of paramount importance, have been cautious in adopting GPT systems. These sectors have observed instances where GPT systems hallucinate or conflate information, even when connected to the internet and capable of linking to sources. For example, a lawyer relying on outputs generated by ChatGPT during a case faced fines and sanctions due to potential inaccuracies.

By front-loading AI systems with relevant data and intervening before the system can produce non-factual information, AI21 has demonstrated a potential solution to the hallucination problem. This breakthrough could lead to mass adoption, particularly in the fintech arena. Traditional financial institutions, traditionally hesitant to embrace GPT technology, could now leverage Contextual Answers to enhance their operations. Moreover, the blockchain and cryptocurrency communities, which have had mixed results with chatbots, may also benefit from the improved trustworthiness and accuracy provided by Contextual Answers.

In summary, AI21 Labs’ Contextual Answers engine represents a significant advancement in the blockchain industry. By addressing the trustworthiness issues associated with AI models and mitigating hallucination problems, this technology opens up new possibilities for businesses across various sectors. With the ability to steer AI answers based on organizational knowledge, Contextual Answers paves the way for widespread adoption of AI in finance, law, and other industries where accuracy is critical.

We will continue to update Phone&Auto; if you have any questions or suggestions, please contact us!


Was this article helpful?

93 out of 132 found this helpful

Discover more


Curve's value increases by 22% as the founder sells CRV tokens to repay an Aave loan.

Curve experiences significant growth, with a 22% increase, thanks to the founder's successful sale of CRV to influent...


Justin Sun's Huobi-linked exchange, WhiteBIT, raises concerns over its lending platform.

WhiteBIT, an exchange associated with Justin Sun, has recently gained attention for offering an impressive Annual Per...


Shady Transactions Raise Eyebrows as $110 Million Evaporates from HECO Bridge and HTX Exchange – What in the Crypto World is Happening?

Recent blockchain breaches on the HECO bridge and HTX platform have been reported by security firms, resulting in an ...


HTX From Hot to Cold - $258 Million Flows out since Resuming Operations Last Month

After a recent security breach, HTX, which is associated with Justin Sun, has experienced a significant $248 million ...


Humorous and Professional Announcement from HTX Exchange

HTX, a popular digital asset exchange, has announced the resumption of deposit and withdrawal services for top crypto...


🚀 Liquid Restaking Tokens: The Hype, The Risks, and The Future of Ethereum DeFi! 🌟

Discover the latest developments in the world of blockchain technology with our weekly newsletter. This week, join Sa...