PulseAugur
EN
LIVE 22:20:03
Русский(RU) Откуда они берут данные? У языковых моделей был интернет. Человечество 30 лет бесплатно писало туда всякую шнягу, так что LLM оставалось просто найти способ, ка

AI learns to describe scents using human-annotated data · 2 sources tracked

AI models are being trained to understand and categorize scents using a standardized perfume lexicon, with human annotators worldwide describing smells. This process aims to bridge the gap between molecular composition and human olfactory perception. The data for training these models is derived from the vast amount of information available on the internet, which has been populated by human contributions over decades. AI

IMPACT This approach could lead to AI systems capable of understanding and classifying complex sensory data like smells, with potential applications in various industries.

RANK_REASON AI model training on human-annotated scent data.

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

AI learns to describe scents using human-annotated data · 2 sources tracked

How we ranked this

Signal score
20 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
AI model training on human-annotated scent data.
Source corroboration
2 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
model release, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [2]

  1. Mastodon — mastodon.social TIER_1 Русский(RU) · [email protected] ·

    About 70 more annotators worldwide are describing scents according to a unified perfume dictionary. And the neural network is learning to match one with the other, for example: this set of molecules in

    Ещё около 70 разметчиков по миру описывают запахи по единому парфюмерному словарю. А нейронка учится сопоставлять одно с другим, типа: вот такой набор молекул в такой концентрации = "пахнет персиком". Сейчас у них уже >10 млн человеческих меток. По дороге, пока они это всё собира…

  2. Mastodon — mastodon.social TIER_1 Русский(RU) · [email protected] ·

    Where do they get their data from? Language models had the internet. Humanity spent 30 years writing all sorts of junk there for free, so LLMs just had to find a way.

    Откуда они берут данные? У языковых моделей был интернет. Человечество 30 лет бесплатно писало туда всякую шнягу, так что LLM оставалось просто найти способ, как туда прийти и всё спарсить-заскрейпить. А у запахов своего интернета нет. По подсчётам фаундера Osmo, до них во всех б…