PulseAugur
LIVE 08:28:55
research · [6 sources] ·
0
research

AI models show dangerous inconsistency in carb counting, study finds

A study involving 26,904 queries across four leading AI models—OpenAI's GPT-5.4, Anthropic's Claude Sonnet 4.6, and two Google Gemini versions—revealed significant inconsistencies in carbohydrate estimations from food images. Despite using the same prompts and images repeatedly, the models produced widely varying results, with some estimations differing by hundreds of grams, potentially leading to dangerous insulin dosing errors for diabetes patients. The study also highlighted instances where models misidentified foods or hallucinated ingredients, further compromising their reliability for health applications. AI

Summary written by gemini-2.5-flash-lite from 6 sources. How we write summaries →

IMPACT Highlights critical reliability issues in AI models for health applications, potentially impacting patient safety and requiring robust validation.

RANK_REASON Research paper detailing AI model inconsistencies with significant implications for health applications.

Read on Mastodon — mastodon.social →

COVERAGE [6]

  1. Mastodon — fosstodon.org TIER_1 · [email protected] ·

    I Asked AI to Count My Carbs 27,000 Times. It Couldn’t Give Me the Same Answer Twice. > Ask ChatGPT to estimate the carbs in your lunch. Now ask it again. And a

    I Asked AI to Count My Carbs 27,000 Times. It Couldn’t Give Me the Same Answer Twice. > Ask ChatGPT to estimate the carbs in your lunch. Now ask it again. And again. Five hundred times. > You’d expect the same answer each time. It’s the same photo, the same model, the same questi…

  2. Mastodon — mastodon.social TIER_1 · hacker_news_bot ·

    📜 Latest Top Story on # HackerNews : He asked AI to count carbs 27000 times. It couldn't give the same answer twice 🔍 Original Story: https://www. diabettech.co

    📜 Latest Top Story on # HackerNews : He asked AI to count carbs 27000 times. It couldn't give the same answer twice 🔍 Original Story: https://www. diabettech.com/i-asked-ai-to-c ount-my-carbs-27000-times-it-couldnt-give-me-the-same-answer-twice/ 👤 Author: sarusso ⭐ Score: 55 💬 Nu…

  3. Mastodon — mastodon.social TIER_1 · CuratedHackerNews ·

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www. diabettech.com/i-asked-ai-to-c ount-my-carbs-27000-times-it-couldnt-

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www. diabettech.com/i-asked-ai-to-c ount-my-carbs-27000-times-it-couldnt-give-me-the-same-answer-twice/ # ai

  4. Mastodon — mastodon.social TIER_1 · h4ckernews ·

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www. diabettech.com/i-asked-ai-to-c ount-my-carbs-27000-times-it-couldnt-

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www. diabettech.com/i-asked-ai-to-c ount-my-carbs-27000-times-it-couldnt-give-me-the-same-answer-twice/ # HackerNews # AI # Carb # Counting # Inconsistency # Diabetech # HealthTech # DataAnalys…

  5. Mastodon — mastodon.social TIER_1 · [email protected] ·

    Letting AI play my game - building an agentic test harness to help play-testing https://blog.jeffschomay.com/letting-ai-play-my-game # HackerNews # Tech # AI

    Letting AI play my game - building an agentic test harness to help play-testing https://blog.jeffschomay.com/letting-ai-play-my-game # HackerNews # Tech # AI

  6. Mastodon — mastodon.social TIER_1 · [email protected] ·

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www.diabettech.com/i-asked-ai-to-count-my-carbs-27000-times-it-couldnt-gi

    He asked AI to count carbs 27000 times. It couldn't give the same answer twice https://www.diabettech.com/i-asked-ai-to-count-my-carbs-27000-times-it-couldnt-give-me-the-same-answer-twice/ # HackerNews # Tech # AI