Andrew Ng: Most Students Use Generative AI for Homework, Direct Answers, or Completing Papers
Andrew Ng stated that using AI for homework harms retention.
He noted that while most students use generative AI for writing assignments, providing direct answers, or completing papers, short-term grades may improve, but long-term memory retention and independent performance after removing the tool will be worse.
He attributed this to "cognitive offloading": the model completes retrieval, organization, reasoning, writing, and error correction for students, who receive the results of task completion without experiencing the extraction, trial and error, explanation, and repeated practice necessary for forming knowledge structures. The apparent ability increase when AI is present exposes capability gaps once AI is removed.
A randomized experiment involving undergraduates found that students allowed to use generative AI performed better on immediate tests, with some gains still present in retention tests a week later, approximately 5.1 percentage points; researchers emphasized that results depend on how students specifically use AI, and not all usage methods will reduce learning effectiveness.
Another group of studies and educational observations found contrasting risks: when students unrestrictedly let ChatGPT replace thinking, the retention test accuracy was 57.5%, compared to 68.5% for traditional learning groups, an 11 percentage point gap; a field study of Chinese high school students found that generative AI improved homework scores by 18% and reduced completion time by 30%, but monthly test scores dropped by 20% after six months.
The differing research results are not contradictory: AI as an explainer, questioner, feedback tool, or personalized practice system may enhance learning efficiency; AI as an answer generator, homework writer, or paper proxy reduces the frequency with which students complete critical cognitive steps themselves. The determinant of learning effectiveness is not "whether AI is present," but whether reasoning, recall, and error correction in tasks are still completed by students.
The most practical use is "do first, then ask": first write out the problem-solving path, code, argument, or summary without looking at AI answers, then ask AI to point out flaws, refute assumptions, generate variant problems, and test oneself with closed-book recitation. For example, when learning calculus, independently derive a problem first, then let AI check each step without directly providing the final solution; the next day, redo similar problems without AI assistance to test whether true capability has been formed.
In market mechanisms, schools, training institutions, and employers are facing the issue of separation between "demonstrable output" and "real ability." Students and knowledge workers will purchase AI tools to improve immediate delivery speed, while educational institutions need to increase oral exams, on-site tasks, process records, stage tests, and tool-free assessments; beneficiaries will be platforms that can design AI as tutors, examiners, and feedback systems, while those under pressure are assignment and certification systems that rely solely on generating final answers without verifying the learning process.
Source: Public Information
ABAB AI Insight
Andrew Ng's judgment is counterintuitive: as a co-founder of Coursera, one of the founding leaders of Google Brain, and a long-time advocate for AI in education, he is not against AI entering education but is distinguishing between "doing work with AI" and "training people with AI." He pointed out that models are very good at completing text, code, and research tasks for people, but this efficiency leads learners to continuously outsource parts that need to be practiced personally; when students encounter similar problems next time, they call upon the model again, thus improving surface output while not accumulating personal transferable skills.
The capital path of educational products will also change accordingly. In the past, online education platforms mainly sold course content, expert explanations, and question banks; with the emergence of generative AI, low-cost explanations and answer generation have quickly become commoditized. The new scarce resource becomes the ability to create effective practice: diagnosing student errors, generating variant problems based on mastery levels, requiring students to answer first, delaying the revelation of hints, tracking reasoning processes, and implementing tool-free assessments at key points. Mechanisms like Duolingo's continuous practice, Khan Academy's conversational tutoring, and traditional spaced repetition systems, the core value is no longer "answering questions," but enabling students to continuously engage in cognitive work that cannot be skipped.
Historically, the introduction of calculators into math classrooms also sparked similar debates: they can eliminate repetitive arithmetic but cannot replace number sense, modeling, and derivation. Search engines improve the efficiency of information retrieval but cannot automatically form argumentative abilities; GPS reduces navigation burdens but may weaken a person's active encoding of spatial routes. The impact of generative AI is stronger because it replaces search, drafting, structuring, language expression, and part of reasoning all at once. Therefore, it is not merely a "stronger calculator" but a cognitive agent that can take over the entire assignment process.
This is a form of technological substitution, but the first to be substituted is not the teacher, but the "necessary friction" in the training process. Long-term memory relies on retrieval practice, spaced review, error correction, and active generation; these steps make learning slower and less comfortable in the moment but constitute the ability to complete tasks after removing the tool. If AI eliminates all friction, short-term indicators will look better, but long-term capabilities will become thinner; if AI specifically retains and strengthens these frictions, it can instead become a scalable personal coach.
ABAB News · Cognitive Laws
AI does the work for you, but it won't leave you with the ability.
Scores reward results, memory rewards struggle.
Tools increase output, practice determines ability.