Teaching AI Agents to Ask Better Questions by Playing "Battleship"

MIT researchers use the classic game as a test bed for AI agents, finding a small model can outperform the biggest ones at 1 percent of the cost.

At Data Tribes, we encountered this compelling piece from MIT News and felt it deserved a closer look, especially for anyone working in AI research, data science, or building intelligent agents.

How AI Threatens the Giants of Consulting

For decades, scale was the moat that protected the giants of consulting. McKinsey, Bain, BCG, and the Big Four Deloitte, EY, KPMG, and PwC relied on armies of junior consultants to take on large, complex projects that smaller firms simply couldn't staff. That advantage is now eroding fast, and AI is the reason.

At Data Tribes, we encountered this compelling piece from the Financial Times and felt it deserved a closer look, especially for anyone working in professional services, data, or AI advisory across the region.

A More Efficient Way to Train Large AI Models Emerges

Researchers have introduced a new AI scaling framework called Item Response Scaling Laws (IRSL) that could reduce computational costs by up to 99 percent. By borrowing concepts from educational testing, the method allows AI developers to predict model performance more efficiently, potentially transforming how large language models are trained in the future.