engblogs

Explores how computation, machine learning, and design theory reshape aesthetic judgment in architecture by translating algorithms into tangible installations and interactive visualizations within the Beyond Data-Driven Aesthetics exhibit.

MIT’s inaugural Music Technology Research Showcase blends AI-driven music research with live performance, from EEG-based neural decoding of imagined tunes to real-time co-improvisation visuals and generative models across interdisciplinary music-technology research.

Explores how computation, algorithmic design, and data-driven AI translate into tangible aesthetics through the MIT exhibition 'Beyond Data-Driven Aesthetics,' using design reconstruction and visualization to interrogate aesthetic judgment across architecture and the applied arts.

MIT's new Music Technology and Computation program marks its inaugural showcase, pairing live performances with cutting-edge AI, EEG-based imagined music decoding, and generative visualization to highlight interdisciplinary research at the intersection of music, engineering, and computation.

Efficient high-cardinality fixed effects estimation for Instacart marketplace experiments using PyFixest and Fixest, powered by the Frisch-Waugh-Lovell theorem and the method of alternating projections.

HP scales the OpenAI Frontier partnership from pilots to enterprise-wide AI, leveraging a unified context and governance platform to deploy AI across customer experiences, security, and productivity workflows.

HP scales OpenAI Frontier from pilots to a governed, production-ready, enterprise-wide AI platform that enhances customer experiences, security, workforce productivity, and software delivery.

HP scales its OpenAI Frontier partnership from pilot projects to a unified, governed enterprise AI platform that enables AI-powered customer experiences, streamlined workflows, and accelerated software development across the organization.

Previewing GPT-5.6 Sol, Terra, and Luna, this technical post highlights a safety-first, multi-tier AI stack with stronger safeguards, enhanced reasoning and ultra mode, and benchmarking gains in coding, biology, and cybersecurity, in a phased rollout with governance-aware considerations and pricing details.

Previewing GPT-5.6 Sol, Terra, and Luna as a safety-forward, multi-tier AI suite with enhanced reasoning, ultra-mode subagents, and strong cybersecurity and coding benchmarks, rolled out in a phased general-availability plan.

Previewing GPT-5.6 Sol: a multi-model rollout (Sol, Terra, Luna) with enhanced safety, performance benchmarks in coding, biology, and cybersecurity, intensive red-teaming, and phased access ahead of broader availability.

Masked IRL leverages LLMs to clarify vague prompts, filter irrelevant details via masking, and align robot trajectories with the user's intent using fewer demonstrations for safer, more efficient manipulation in real-world settings.