Skip to content
Trending storyDeveloping

Ai2 reports GPT-6 models over-help students less than earlier OpenAI models

1 article1 sourcesince Oct 12Last article 1h ago ·

Overview

AISummary of 1 article

Ai2 says in tests that GPT-6 models over-help students less than earlier OpenAI models.

Ai2 also reports that the GPT-6 models score higher overall when explicitly instructed to offer help where needed and leave students to work through parts of a problem they can handle themselves.

The report is a single post on X from Ai2 (@allen_ai). It gives no test details, sample sizes, or specific scores, so the size of the difference is not stated.

Written by AI from the articles below · updated Oct 12, 4:52 PM ET

Check the sources:

Latest developmentOct 12, 4:43 PM ET
GPT-6 models over-help less than earlier OpenAI models in tests

Article timeline

The articles in this story. Times are ET.

Oct 12
  1. Ai2Official
    GPT-6 models over-help less than earlier OpenAI models in tests

    AIIn these tests, GPT-6 models over-help students less than earlier OpenAI models do. They also score higher overall when explicitly instructed to offer help where needed and leave students to work through parts they can handle themselves.

    Image from @allen_ai's post

Heat trend

Not enough continuous observations to show a trend yet.