The tests that grade AI may be getting it wrong

Дата публикации: 30-09-2026 13:20:14

Before a new AI model reaches the public, its developers run it through a battery of tests known as "benchmarks," which score it on everything from reasoning ability to how safe it is for people to use. Billions of investment dollars ride on these benchmark scores, and policymakers increasingly cite them to shape regulations and government procurement decisions that will guide the development of AI.

Основное содержимое страницы с новостью.

Схожие новости

#	Наименование новости	Тональность	Информативность	Дата публикации
1	Anthropic and OpenAI sound the alarm on AI safety and seek to shape how it’s controlled	0	7.76	28-09-2026
2	Q&A: Expert says people often confuse behavior with intention with AI	0	7.85	30-09-2026
3	Бизнес стал внимательнее просчитывать жизненный цикл ИИ-проектов после запуска	0	14.38	28-09-2026
4	В тумане метрик. Получится ли у государства измерить эффект от внедрения ИИ	0	15.49	01-10-2026
5	Even AI can’t perfectly assemble Ikea furniture—yet	0	12.23	28-09-2026
6	Google restricts access to new AI model over safety concerns	0	10.33	01-10-2026
7	Microsoft drafts ‘humanist’ AI code of conduct amid safety debate	0	6.96	15-09-2026
8	OpenAI scraps release of new AI model over safety concerns — WSJ	0	6.81	29-09-2026
9	OpenAI hack sparks further concern over AI models going rogue	0	7.01	28-09-2026

Классификация: Наука. Схожих патентов: 0. Схожих новостей: 9. Тональность: 0. Информативность: 6.19. Источник: techxplore.com.