What's Hot in AI
With so much information out there, it’s hard to know what to read first. NCMA’s 'What's Hot' highlights our top recommended articles to stay abreast of recent developments.
Updated July 22, 2025
Scale AI Layoffs Underscore Failure to Operationalize Growth—Even Among AI Leaders
Scale AI, a major provider of data-labeling services to firms like OpenAI and Google, has laid off 14% of its full-time staff and ended contracts with 500 global contractors. The move comes just weeks after Meta acquired a 49% stake in the company and brought founder Alexandr Wang into its executive ranks. Despite being at the center of the generative AI boom, Scale admitted it expanded too quickly, creating duplicative teams and unclear priorities. The layoffs are part of a restructuring to consolidate its GenAI efforts into five product verticals, but the sudden reversal highlights how even AI-first companies are struggling to scale in operationally sound ways.
CDAO Partners with Frontier AI Firms to Advance National Security Mission
The Department of Defense’s Chief Digital and Artificial Intelligence Office (CDAO) has formalized partnerships with Anthropic, Google, OpenAI, and xAI through multiple-award contracts worth up to $200 million each. These agreements aim to provide secure access to advanced AI foundation models and mark a shift from experimentation to operational use in defense and national security. For contracting professionals, these awards highlight the need for flexible acquisition frameworks that address model access, data security, performance accountability, and IP rights—especially when working with non-traditional vendors. The scale of these deals signals a new, more urgent phase in AI procurement.
Mistral Unveils Major Feature Update for Le Chat: Deep Research, Voice, Image Tools
Mistral AI has introduced major updates to its chatbot Le Chat, adding Deep Research for structured reports, real-time Voice Mode (powered by Voxtral), Think Mode for advanced reasoning, a redesigned image editor, and Projects for managing multi-threaded workspaces. These features position Le Chat as a serious alternative to tools like ChatGPT and Gemini, offering richer functionality for both consumers and enterprise users. For contracting managers, this reflects how smaller AI vendors are rapidly closing capability gaps. As agencies explore these tools, contracts must address performance, privacy, accessibility, and long-term support for voice, multimodal, and integrated workspace features.
A Practical Framework for Deciding What AI Should—and Shouldn’t—Do
A new article in California Management Review introduces the EPOCH Methodology to help organizations assess whether AI should automate or augment a task. EPOCH scores work across five human dimensions: Empathy, Presence, Opinion, Creativity, and Hope. Based on these, it outlines four strategies—Status Quo, Augmentation-Led, Human-in-the-Loop, and Displacement-Driven—to guide implementation. For contracting managers, EPOCH offers a disciplined way to define AI requirements. Instead of defaulting to vendor-led solutions, acquisition professionals should assess task characteristics with customers, select the right strategy, and align contracts accordingly. This ensures AI investments support mission needs and include clear goals, reassessment points, and human-AI role definitions.
Reverse Thinking Gets a Boost from AI—and a Role in Strategy
A new California Management Review article explores reverse thinking—flipping assumptions to spark innovation and reveal blind spots. It defines three types: assumption reversal (challenging norms), role reversal (seeing from an outsider’s view), and tactical reversal (doing the opposite of convention). Examples include Shiseido’s inverted product design and Forus Health’s portable diagnostics. The authors highlight how AI tools, especially generative models and graph systems, can scale this mindset. For contracting managers, reverse thinking offers a way to improve market research, stress-test requirements, and explore nontraditional solutions. Using AI early in planning can surface overlooked risks and alternatives, improving acquisition outcomes.