NEW YORK, NY (May 10, 2024) - The first inclusive Large Language Model (LLM), Latimer.ai, announced the release of a comprehensive white paper that addresses the pressing challenges of bias in LLMs. Titled "Bias Neutralization Framework: Measuring Fairness in Large Language Models with Bias Intelligence Quotient (BiQ)," the paper presents a pioneering structure called Bias Intelligence Quotient (BiQ) for the detection and mitigation of bias, setting a new standard in the industry for responsible AI development. The release of the paper further solidifies Latimer.ai’s commitment to accurate historical data, and has created a clearly defined pathway to building  innovative tools that will identify and measure bias in unique ways and ultimately highlight corrective solutions for users. 

"Generative AI and LLMs hold incredible potential to influence society positively, but only if we can trust it to act fairly," said John Pasmore, CEO of Latimer. "This white paper reflects our commitment to leading the charge in ethical AI, by ensuring LLMs are trained on diverse datasets and that our technologies are built not only with intelligence, but with integrity. We believe that AI should be transparent and everyone should be aware of what bias an LLM displays in its interactions."

Latimer.ai’s release of the white paper is a part of their broader initiative to foster transparency and responsibility in AI. The paper outlines a detailed methodology for organizations to assess and address various forms of bias, ensuring LLMs perform ethically across different demographics and scenarios. This is crucial as AI becomes increasingly integrated into critical areas such as healthcare, finance, law enforcement and other areas where biases can have significant adverse effects.

“The non-deterministic nature of LLMs makes  them exceptionally powerful, but also renders them susceptible to biases. Bias is introduced in multiple ways within LLMs, including through data, human interactions, and algorithms. Understanding and mitigating the various types of biases is not just a technical challenge—it's a fundamental aspect of developing AI that truly benefits all," adds Latimer.ai Technical Advisor, Malur Narayan. "Bias Intelligence Quotient (BiQ) is a comprehensive metric that allows organizations to measure AI bias in LLMs systematically and apply mitigations effectively. It’s about creating technology that mirrors the diversity and richness of human experience.”

Narayan led the development of the paper with Pasmore, along with additional contributions from engineer Elton Sampaio and AI thought leader, Vijay Raghavan, PhD. Morgan State University's Gabriella Waters, also contributed to the paper, further extending the established partnership with Latimer.ai, wherein the university's Center for Equitable AI and Machine Learning Systems (CEAMLS) is already underway to test and validate Latimer's performance and functionality while inviting all Morgan State students to use the LLM. 

The white paper was published on arXiv, a leading distribution service and open-access archive for nearly 2.4 million scholarly articles in the fields of physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, electrical engineering and systems science, and economics. 

Consumers and businesses looking to experience the new platform can join Latimer.ai at www.latimer.ai. 

ABOUT LATIMER 

Artificial Intelligence (AI) is amplifying the bias, inaccuracy, and erasure of Black and Brown cultural data. Latimer was built to solve this. Latimer has gathered some of the most notable names across various industries to create a more inclusive and equitable Large Language Model (LLM) platform, utilizing a RAG Model with its proprietary vector database that sits on top of a foundation model, which is currently Open AI's Chat GPT. Named for Lewis Latimer, a Black inventor whose legacy and historical scientific contributions are often overlooked, the LLM provides users with more accurate details, reflecting the experience, culture, and history of Black and Brown people. Latimer was founded and is led by serial entrepreneur John N. Pasmore, who also established Latimer’s parent company, FutureSum.AI, which works to build premier AI tools to improve how organizations engage with diverse audiences. John was recently accepted as one of five Black innovators into the 2024 class of the Lewis Latimer Fellowship Program. To learn more, please visit www.latimer.ai.