{
  "id": 13970982,
  "title": "Should an AI Tutor Just Give You the Answer? I Tested It",
  "url": "https://urgent.news/2026/10/12/should-an-ai-tutor-just-give-you-the-answer-i-tested-it",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-12T14:00:00.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/vikrantnegi/should-an-ai-tutor-just-give-you-the-answer-i-tested-it-2ldh"
  },
  "original_language": "en",
  "account": "In a final project for the 100x Engineers AI course, the author built an AI tutor for SQL that didn't immediately provide answers. The tutor was then compared to a normal version that gives answers right away. Six people practiced with one version or the other, and a week later took a 12-question SQL test. The \"try first\" group performed better overall, with all 36 of their answers correct on the later test. However, when asked why they didn't claim the victory, the author explained that two of their ten checks failed. The results were not as different as they had planned. While the \"try first\" group did make more attempts before seeking help, they didn't perform better on the first try. Additionally, the test was conducted at the author's leisure, not in real-time, which could have influenced the results. The author concludes that while making learners try first may change their behavior after a wrong answer, it doesn't necessarily improve their overall performance.",
  "summary": "I built an AI tutor for SQL that won't hand you the answer straight away. Then I tested it against a normal version that does. Six people practised with one version or the other. A week later, they took a 12-question SQL test with no tutor at all. The \"try first\" group got all of their answers right: 36 of 36 (3 people × 12 questions). The \"give me the answer\" group got 25 of 36. That looks like…",
  "key_points": [
    "\"Try first\" AI tutor group performed better overall in SQL test",
    "Two of ten checks failed, limiting conclusive results",
    "Test conducted at author's leisure, not real-time"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}