Recently, Sentient Labs, led by Peter Thiel's Founders Fund, launched a new open-source AI search framework—Open Deep Search (ODS)—with a seed funding of up to $85 million. This framework aims to provide AI with capabilities for search, reasoning, and verification, mitigating the issue of AI hallucinations.
What is AI hallucination?
AI hallucination refers to the phenomenon where AI models generate information that seems reasonable but is actually incorrect. For example:
Fabricating non-existent papers or citations
Confusing facts, causal relationships, or timelines
Piecing together seemingly credible but actually erroneous conclusions
The fundamental reason for this phenomenon lies in the fact that current AI models mainly rely on pattern recognition in training data rather than truly understanding and verifying the authenticity of information.
ODS: AI's fact-checking assistant
ODS is an open-source search proxy system that enables multi-tool collaboration, aimed at providing AI models with capabilities for search, reasoning, and verification. Its core components include:
Open Search Tool (OST)
OST can understand user intent, intelligently generate search terms, deeply crawl the internet for effective information, and perform semantic rearrangement, filtering, and aggregation to improve the quality and relevance of search results.
Open Reasoning Agent (ORA)
ORA simulates human multi-step reasoning processes and can actively perform secondary queries when information is insufficient, invoking various external tools or plugins, and even generating and executing Python code to solve complex logical or computational needs.
Advantages of ODS
Interpretability: Every operation of ODS is visible, allowing users to trace back the AI's reasoning chain and information sources, thereby enhancing the system's transparency and credibility.
Customizability: ODS supports the integration of any large language model and external tools or plugins, allowing users to freely customize reasoning rules according to their needs, catering to different application scenarios.
Reducing misinformation: By cross-referencing multiple information sources and actively performing secondary queries, ODS effectively reduces the spread of misinformation, false information, and misleading information by avoiding incorrect conclusions drawn solely from keyword matching.
Practical application examples
Healthcare sector: AI models may generate incorrect diagnostic suggestions, leading to serious consequences. By integrating ODS, healthcare AI systems can automatically search for the latest medical research and authoritative guidelines before generating diagnostic suggestions, verifying the accuracy of information and thus improving the reliability of diagnoses.
Financial sector: AI models may make investment suggestions based on outdated or erroneous data. ODS can help financial AI systems obtain the latest market data and analysis reports in real-time, conduct multi-party verifications, and ensure the accuracy and timeliness of investment advice.
Summary
The launch of ODS marks a significant breakthrough in open-source AI search technology. It not only enhances the transparency and controllability of AI systems but also provides developers with powerful tools to build more reliable and trustworthy AI applications. With the continuous development of ODS, we have reason to believe that open-source AI will play a more important role in future technological ecosystems.
This article is for sharing and communication purposes only and does not constitute investment advice.
