NYT Cooking Tests AI-Powered Recipe Tweaks, Sparking User Debate

The New York Times has begun a gradual rollout of an AI‑driven modification layer for its NYT Cooking iOS app. The tool lets subscribers ask for ingredient substitutions, alternative cookware, or any serving‑size adjustment, and it rewrites the recipe steps accordingly.

In a demo, a user requested baby spinach instead of kale; the assistant noted that spinach wilts faster and recommended adding it later in the cooking process. All customized versions remain private and cannot be shared publicly.

Feedback in the app’s comment section has been polarized. Some readers welcome the convenience, while others criticize the environmental cost of running large language models and the inability to disable the feature. The Times says the chat interface is optional, but there is currently no off‑switch.

Why it matters for GPU / AI infrastructure

Features like this rely on inference‑heavy models that run continuously on GPU clusters. As consumer‑facing apps embed generative AI, demand for low‑latency, high‑throughput GPU capacity grows — reinforcing the need for scalable, cost‑effective cloud GPU solutions such as those offered by aigpu.

The episode illustrates how quickly AI is moving from backend experimentation to front‑line product experiences, and why infrastructure providers must stay ahead of inference workload spikes.