The human engine of artificial intelligence: datawork and LLM performance
Adriana Alvarado of IBM argues that datawork – the human processes of collecting, curating and documenting datasets – is the core driver of artificial intelligence performance and fairness. Her talk explores why Large Language Models amplify the stakes of these human choices and how synthetic data and robust governance fit into the solution.
AMD MI300X powers training of Zyphra’s ZAYA1 Artificial Intelligence model
Zyphra trained ZAYA1, a large-scale mixture-of-experts foundation model, using AMD Instinct MI300X GPUs, AMD Pensando networking, and the AMD ROCm software stack. Zyphra reports the model matches or exceeds leading open models on reasoning, mathematics, and coding benchmarks.
What’s Grokipedia, Musk’s Artificial Intelligence-powered rival to Wikipedia?
Elon Musk launched Grokipedia, an Artificial Intelligence-powered site that largely reproduces Wikipedia content but often lacks citations and transparency. PolitiFact found copied text, missing or incorrect references and unclear correction processes.
The business side of artificial intelligence content creation
Generative artificial intelligence is shifting marketing relationships from vendors to partners and elevating contracts as strategic assets. New deal structures, from archive-driven fan remixes to omniverse production and feed-level integration, are reshaping how brands create and monetise content.