Exploring the frontiers of AI in business through rigorous research and practical applications
SPARK R&D is classified into six categories
The hardware and software technologies underlying modern AI
AI's impact on corporate strategy, organization, company culture and change management
AI's impact on civic leadership, organization and public service
AI's impacts on productivity, personalization, and privacy
Micro- and Macro-economic value, workforce opportunity, disruption and displacement
The science of incentivizing, controlling, and governing AI
Explore our ongoing research initiatives and collaborative projects.
As artificial intelligence systems grow in scale, complexity, and capability, the question of how these systems can be effectively governed is increasingly urgent. This project is developing a cross-sector analysis of how industry and public sector leaders are confronting the challenges of developing effective governance policies and ensuring accountability in frontier AI development.
This project developed and tested a novel experimental system to watch and detect anomalies in blockchain network data flows, using realistic blockchain network simulations and machine learning trained on data collected from those simulations. The test system itself is user-configurable and interpretable.
While the 1980s demanded defensive coding and the 2010s formalized data quality management practices, agentic AI requires a new paradigm. This project introduces Zero Trust Data Quality (ZTDQ), a framework that applies zero trust principles to data quality assessment.
This project examined and defined Explainability Driven Data Quality (EDDQ), a methodology that leverages machine learning explainability weights to automate most data quality management decisions for machine learning models.
Forum & ARC Sponsors OnlyWith scalable architectures, improved tooling, the maturation of standards such as RDF, OWL, SHACL, and visualization tools such as knowledge graphs, semantic technologies are no longer exclusively the domain of specialized academic research, subject matter experts (SMEs), and selected companies. This project examines how generalized semantics data engineering and analytic frameworks are being applied in pharmaceutical drug discovery and development.
Forum & ARC Sponsors OnlyA major challenge in evaluating a company's readiness for an AI-driven IT architecture is understanding the complex interplay between advanced data processing, analytics capabilities, and intelligent storage technologies. This project focuses on how companies can effectively leverage existing legacy systems while rigorously evaluating the performance, cost, and policy implications of next-generation AI technologies.
Forum & ARC Sponsors OnlyAgentic AI has fueled expectations of rapid productivity improvements and significant economic impact. Empirical studies show early-stage micro-level gains, such as faster task execution, fewer errors, and more consistent outputs. However, these micro-level improvements have not translated into measurable macroeconomic effects. Why?
Every major information-based technology in the last half-century has displaced legacy tasks, reorganized firms, and changed skill premiums. In the 1970s, relational databases and transaction processing drove large-scale automation of clerical, accounting, inventory, payroll, and logistics work, triggering major displacements of white-collar workers. Today, there is little new or novel in the contention that AI will trigger workforce displacement or materially adjust skill premiums. A potentially more important question is how soon, and in what ways, will fully automated systems take hold in higher skill-level business functions, with no human effectively in the loop.
Forecasts of AI progress are numerous, yet there is little consensus on measurement or prediction. This project examines prediction markets as a mechanism for aggregating diverse projections about AI trajectories. Special attention will be given to the composition of market participants—domain experts, generalists, and financially motivated forecasters—to address how incentives shape prediction behavior. The study aims to clarify the conditions under which prediction markets provide calibrated insights into AI progress and uncertainty.
Interested in partnering on research or accessing our findings? Join our consortium to collaborate on cutting-edge AI research.
Become a Research Partner