Google put a study coach inside Search. The useful question is who checks the answers.

What is genuinely new here

Two of these features are more interesting than the headline suggests. The first is the diagnostic quiz that precedes the study plan. Asking a student what they already know before deciding what to teach them is standard practice in tutoring and almost absent from study apps, which usually start from a syllabus rather than from the person.^1,3

The second is generating a visual for a concept on demand. A pH scale, a DNA structure you can turn. For a student who does not have a teacher available at ten at night, an illustration produced in response to their specific confusion is a real improvement over searching for a diagram drawn for somebody else's question.^1,2

The rest is more familiar. Flashcards, notebooks and quiz generation exist in a dozen products already, and the meaningful change is that they now sit where students already are rather than in an app they have to choose.^1,3

The problem nobody has solved

A generated practice question can be wrong. So can a generated explanation, a generated diagram and a generated answer key. This matters more in study tools than almost anywhere else, because the entire purpose is to be used by someone who does not yet know enough to catch the error.^1,2

Google's announcement does not describe how question accuracy is verified, and no independent evaluation was available when we published. That is not an accusation. It is the single most important unanswered question about a product of this shape, and it deserves to be stated as an open item rather than left out of the summary.¹

The standardised exam claim raises it sharply. A practice question for the MCAT or the LSAT that is subtly wrong does not merely fail to help. It teaches a confident mistake to someone preparing for a test that shapes a career.^1,2

How to use them without being taught something false

The tools are worth using. They are not worth trusting in the way a textbook can be trusted, and the difference is a habit rather than a restriction.¹

• Use generated quizzes to find gaps, not to confirm knowledge. A wrong answer that reveals confusion is useful even if the question was flawed.
• Check anything surprising against the course material. If a generated answer contradicts the textbook, the textbook wins until a teacher says otherwise.
• Treat generated diagrams as illustrations of an idea, not as sources. They are drawn to be clear, not to be correct in detail.
• For standardised exams, use official practice material as the reference and generated questions only for extra volume.
• If a student is using these daily, ask them occasionally to explain an answer rather than show it. That surfaces a misunderstanding faster than any score.

What studying with a machine changes

There is a genuine finding in education research that a struggle a student works through is remembered better than an explanation delivered smoothly. Difficulty is not always friction to be removed. Some of it is the mechanism by which the learning happens, and a tool that answers instantly is very good at removing exactly that.^1,3

This is not an argument against the tools, and it would be a poor one. A student stuck at ten at night with nobody to ask is not experiencing productive difficulty. They are experiencing the absence of a teacher, and a generated explanation is plainly better than giving up. The distinction is between being stuck and being unwilling to start.^1,2

The practical version of that distinction is small. Attempt first, ask second. It costs a few minutes and it is the whole difference between a study aid and a way of never quite learning something.¹

Reading the free year properly

A free year of a paid tier for college students is a straightforward customer acquisition move, and there is nothing wrong with that as long as it is read as one. The value lands during the year. The decision lands at the end of it, when the student has built their study habits on a product that then has a price.^1,3

The offer also splits by geography. American college students get AI Pro with 5TB of storage. Students elsewhere get AI Plus with 400GB. Same announcement, materially different offer, and the difference is not explained.¹

The Classroom side deserves a separate note. Google is giving teachers tools to customise what AI students can use and to track progress, which places a teacher between the student and the model. Given the accuracy question above, that is the most reassuring part of the announcement, and the part least likely to reach anyone studying alone at home.^1,3