AI boosted homework scores, then exam scores dropped: study
A new study reveals that while AI helps students complete homework faster and achieve higher scores, it significantly degrades their performance on exams. This paradox sparks a heated debate on Hacker News about the fundamental purpose of education and AI's role in fostering genuine learning versus enabling superficial completion. Commenters wrestle with whether AI merely exposes existing pedagogical flaws or if it fundamentally alters how a generation will acquire knowledge and critical thinking skills.
The Lowdown
A recent study, highlighted by The Economist, investigates the impact of AI usage on student performance among 27,000 Chinese pupils aged 12-18. The findings present a concerning disconnect: students utilizing AI for their homework showed improved assignment scores and reduced completion times, yet their exam results significantly suffered compared to their non-AI-using peers. This suggests that AI, while efficient for task completion, may be hindering the deeper learning necessary for true comprehension.
- The study involved 27,000 students aged 12-18 in China.
- Roughly 80% of these students used AI tools (like Doubao and DeepSeek) for homework.
- AI users saw their average homework scores increase by 18%.
- Homework completion time for AI users decreased from 64 minutes to 45 minutes.
- Despite these homework boosts, AI-using students scored 20% lower on subsequent exams compared to a control group.
- The research indicates that AI facilitates outsourcing homework rather than deep engagement, leading to a superficial mastery that doesn't translate to exam performance.
This outcome challenges traditional educational metrics, as high homework scores no longer reliably predict exam success when AI is involved, urging a reevaluation of learning methodologies in the age of generative AI.
The Gossip
AI: An Amplifier of Abilities
Many commenters propose that AI acts as an amplifier, enhancing the capabilities of both dedicated learners and those seeking shortcuts. This leads to a "smarter get smarter, dumber get dumber" phenomenon, where motivated students might use AI for deeper understanding, while others use it to bypass learning entirely. Critics, however, warn that AI primarily adds noise, making it harder to discern genuine talent amidst AI-generated output.
Pedagogical Predicaments Exposed
A significant portion of the discussion argues that AI doesn't create new problems but rather exposes deep-seated flaws in the existing education system. Commenters criticize the emphasis on grades over true learning, the nature of homework, and the system's role in stratifying students. Suggestions for reform include making homework optional, implementing randomized exams with unlimited retakes, or adopting a "flipped classroom" model where learning happens outside class and application within. However, others caution these changes might overburden educators or be unsuitable for younger students.
The Homework-Exam Conundrum
The core study finding ignites a debate about the distinct purposes of homework and exams. Many assert that homework's value lies in the 'journey' of practice and understanding, while exams assess true knowledge. AI's ability to short-circuit the homework process bypasses this crucial learning journey. Commenters deliberate whether memorization is essential for mastery, contrasting it with 'learning how to learn,' and note that real-world professional tasks often resemble 'exams' requiring deep, ready knowledge.
New Tool, Old Debates
Several commenters draw parallels between the integration of AI and historical technological shifts, most notably the introduction of calculators. They argue that education must adapt to ubiquitous tools, but express concern that AI, unlike calculators, can bypass more fundamental cognitive development. This raises fears of cognitive decline or a future generation lacking critical thinking and problem-solving skills without AI assistance, mirroring past anxieties about reading or literacy.