Author Archives: Ben Lorica

Pinpointing Arbitrary Claims with Semantic Entropy Confabulation in AI describes the phenomenon where large language models (LLMs) generate fluent yet factually incorrect and arbitrary statements. These erroneous outputs are often sensitive to seemingly irrelevant details, such as the random seed used during generation. For example, an LLM might produce different answers to the same medicalContinue reading “Detecting LLM Confabulations”

Refusal in language models refers to the ability of these models to decline generating responses to harmful, unethical, or inappropriate prompts. This behavior is crucial for maintaining the safety and responsibility of AI systems. It ensures that AI applications do not produce harmful content, perpetuate biases, or engage in unethical behavior. For instance, refusal mechanismsContinue reading “Improving LLM Reliability & Safety by Mastering Refusal Vectors”

How Machines Are Transforming Proofs My career began in mathematics, followed by a transition to quantitative finance. In the early 2000s, I became a data scientist. Mathematics is highly specialized, making it difficult to stay abreast of the latest developments. Therefore, I am far from an expert on the use of AI in mathematics. However,Continue reading “Lessons From The AI Revolution in Mathematics”

Large language models (LLMs) have revolutionized AI application development, but they come with significant challenges. Chief among these is the tendency of these models to produce plausible but false information. The common term for this phenomenon, “hallucinations,” doesn’t fully capture the nature of these inaccuracies. Another crucial aspect of AI development is the evaluation ofContinue reading “BS, Not Hallucinations: Rethinking AI Inaccuracies and Model Evaluation”

By Ben Lorica and Prashanth Rao. Since we released a vector database index nearly two years ago, the landscape of vector search and databases has evolved dramatically. The rise of Retrieval-Augmented Generation (RAG) has been a pivotal factor, with embeddings emerging as the lingua franca of Generative AI. This paradigm shift has spurred a surge inContinue reading “Choosing the Right Vector Search System”

My first job after academia was in quantitative finance, a field that relies heavily on the use of mathematical models and statistical methods to analyze financial markets. One of the most widely used tools in this field is the Fama-French factors, a set of variables developed by Nobel laureate Eugene Fama and Kenneth French toContinue reading “Lessons from the ‘Noisy Factors’ Study”

Apple has unveiled a suite of AI-powered features and enhancements, branded under the umbrella of Apple Intelligence, that integrate seamlessly into iPhone, iPad, and Mac. These innovations include advanced language tools for rewriting, proofreading, and summarizing text, as well as significant updates to Siri, which now offers richer language understanding and contextual awareness. Users canContinue reading “AI at WWDC 2024”