Artificial intelligence

Developers spend most time understanding code, not building it. Katie McDonald explains agentic coding, legacy code modernization, and how AI accelerates software development workflows.

Ever watched a UFC fight and thought, “Wait—what just happened?” This episode of Smart Talks with IBM goes inside the Octagon with Alon Cohen, the tech leader behind UFC’s analytics, to show how AI helps uncover the moments that decide a fight.

How did search evolve into agentic AI? Sam Anthony explains RAG's evolution, from simple retrieval to adaptive systems powered by LLMs. Learn how semantic search, hybrid retrieval, and AI agents enable multi-step research, decision-making, and synthesis.

Right now, an AI agent is booking flights, writing code, and has no idea what it did five minutes ago. Bri Kopecki explains why AI agents need an operating system to manage memory, tools, identity, and guardrails. Learn how Agent OS infrastructure makes AI reliable and scalable.

Agentic AI failures aren’t random—they happen for clear, predictable reasons. Meenakshi Kodati explains the top failure modes in agentic AI systems, from infinite loops to hallucinated planning and unsafe tool use. Learn how better design, constraints, and monitoring improve reliability.

New research makes the case for selective AI sovereignty—that is, prioritizing control over the most critical parts of the tech stack, not every part.

New research suggests that human judgment can come into question when it relies too heavily on AI. In experiments, incorrect AI advice made participants less accurate, more confident and far less likely to say “I don’t know.”

TryAI gave four frontier AI models a colored-pencil toolset and asked them to reproduce the Mona Lisa and Starry Night. Every one of them peaked mid-run and finished worse.

AI just cracked a math problem unsolved since 1939. Experts explain what this means for scientists, engineers—and your job.

Open-weight models let companies control and customize their AI. IBM’s vLLM-Hook uses that openness to debug and fix LLMs from within.

AI coding tools are putting senior software engineers back behind the keyboard after years spent managing teams and designing systems.

What is Inkling? Thinking Machines’ new open-weight AI model prioritizes developer customization and fine-tuning over raw benchmark performance.

IBM Research leader David Cox says generative AI is reshaping how scientists work, accelerating everything from coding to breakthrough research.

From excessive dream sequences to overuse of “tapestry,” AI-generated stories have certain quantifiable idiosyncrasies, says a new study. Can you tell the difference?

AI ”actress” Tilly Norwood is set to star in Misaligned, a feature film blending generative AI, motion capture and real-time inference to create one consistent character. Here’s the engineering challenge behind pulling off 90 minutes of synthetic performance.