Jev, Prolog, Pi, and the dream of probabilistic logic programming (deepclause.substack.com)

Jev is making waves in the AI/ML community with its integration into the latest versions of DeepClause and its Pi extension, enabling probabilistic logic programming that leverages logical predicates in DML/Prolog. This development offers a significant advancement by improving speed and determinism, particularly in applications that would otherwise rely on expensive or unpredictable LLM/Agent approaches. Jev's capabilities are anticipated to enhance SOP2AGENT-style implementations, potentially revitalizing interest in probabilistic logic programming, a concept that has faced challenges in practical adoption for decades. The technical integration allows developers to harness Jev to create sophisticated decision-making frameworks using DML code, enabling nuanced responses to user messages based on probabilistic assessments. By employing predicates to evaluate states, decisions can be made with calibrated probabilities derived from Jev's extensive training data, addressing a historical challenge of reliably sourcing these probabilities from domain experts. This convergence of deep learning and logic programming could open new avenues for building intelligent systems, suggesting that the enduring dream of effective probabilistic logic programming might finally be within reach.