Články z oblasti vzdělávání a poradenství

Agentic AI fails on complex tasks: Patronus bets on “live” simulators

Patronus AI argues agents struggle with real work: interruptions, context shifts, and layered…

Zoom claims a record on a tough AI exam; critics call it an ensemble of others’ models

Zoom reported a 48.1% score on Humanity’s Last Exam. Critics argue it reflects orchestration of…

JPMorgan scaled internal AI without mandates: connectivity to systems made it work

JPMorgan says its internal assistants spread organically: hundreds of thousands of users followed…

Nvidia opens Nemotron 3: models and data as a hedge against cloud dependence

Nvidia released open Nemotron 3 models and included training data and tuning tools. For enterprises,…

Agentic AI hits architecture: design patterns aim to stop expensive pilots

The piece argues agentic AI fails more on system design than on model choice: routing, guardrails,…

AI is both weapon and shield—governance decides which

Deloitte outlines the paradox: AI accelerates attacks (deepfakes, social engineering) but also…

Agents fail when they just assist broken processes

Deloitte warns agentic AI often hits reality: companies automate poorly designed workflows. Success…

Agentic AI could cut drug development time 20–30% - with process redesign

McKinsey says agentic AI could reduce time and costs in drug development by tens of percent. The…

McKinsey: AI speeds up coding—trust, review, and compliance become the bottleneck

Sonar CEO Tariq Shaukat says AI shifted the problem from writing code to reviewing it. Companies…

Listing 55 to 63 out of 114