Google is giving South African university students a year of its paid Gemini plan for free, while simultaneously making Gemini capable of doing considerably more of the work that surrounds studying. It can ingest lecture material, test what a student knows, decide where the gaps are, produce lessons and quizzes to address them, help organise assessment dates and increasingly participate in research. The offer saves students money, but Google is also making a fairly ambitious pitch about where AI should sit between a student and their education.
Eligible students outside the US can claim 12 months of Google AI Plus at no charge until 31 December 2026. The offer extends across more than 140 markets where the plan is available, including South Africa, and includes Gemini Omni, higher Gemini usage limits and 400GB of storage. Students need to provide a payment method and the subscription switches to the normal paid rate after the free period unless they cancel it. Google’s announcement also includes a collection of education features, several of which will be available to people who aren’t paying for AI Plus.
Cost is a real constraint in South Africa because another monthly digital subscription isn’t an inconsequential purchase for many students. Removing it for a year means more people can use the paid version of Gemini without having to decide whether an AI subscription deserves money that could be going towards data, transport or ordinary living costs. It doesn’t solve the larger access problem, since useful AI still assumes a decent device and enough connectivity to make regular cloud-based use practical, but free is materially different from merely available.
Google could have stopped there and had a perfectly serviceable student promotion. Instead, it has been building Gemini around some of the decisions involved in learning itself.
Its student hub includes study notebooks that can work from lecture notes and course material supplied by the student. Gemini can use a diagnostic quiz to identify areas where it thinks the student is struggling and then generate lessons and further questions around those areas. A progress dashboard records how the student is doing, while another forthcoming feature will be able to read a syllabus, extract examination and assignment dates and add them to Google Calendar with permission.
Gemini can also create interactive visualisations, including manipulable 3D models for some subjects, while Deep Research is being incorporated into Gemini Live so that a student can ask the system to investigate a topic and then discuss the resulting research with it. Taken together, these tools make Gemini less like the chatbot someone opens for an occasional explanation and more like software Google expects to remain open throughout the process of studying.
Some of that sounds genuinely useful because some of it is genuinely useful. A student who doesn’t understand a concept can ask for another explanation without having to wait until the next tutorial. Someone balancing a degree with work can hand some of the administrative burden of organising deadlines to software. Lecture notes that have become an indecipherable pile halfway through a semester can be turned into something searchable, testable and easier to revisit.
The harder questions begin once the same system is involved in deciding what the student doesn’t know, producing the material intended to correct that weakness and assessing the answers afterwards. None of those individual functions is especially alarming on its own. Combined, they shift a surprising amount of judgement about the learning process towards a product whose recommendations the student has little ability to inspect beyond whether they appear plausible.
That arrives while universities are still working out which parts of learning should be delegated to generative AI in the first place. The University of Cape Town’s AI in Education Framework acknowledges the everyday presence of systems such as Gemini, ChatGPT and Claude among students and staff, while recognising that they create difficult questions around teaching, assessment and academic integrity. Stellenbosch University’s guidance similarly accepts that students need experience using generative AI because they are likely to encounter it after graduation, while requiring transparency about its role in assessed work.
There is an awkward difference in speed here. Universities have to think about what graduates should know, what an assessment actually measures and how academic integrity works when software can produce competent answers in seconds. Google can ship another Gemini feature.
Students aren’t waiting for institutional consensus either. They already use generative AI to explain concepts, summarise readings, brainstorm essays, check writing and prepare for tests. Google’s opportunity is therefore less about persuading students that AI belongs at university than persuading them that Gemini should be the particular AI through which a growing share of that work happens.
That sits neatly alongside products Google already has inside student life. A student might receive coursework through Gmail, write in Docs, keep material in Drive and organise classes or deadlines through Calendar. Gemini’s advantage is that it can increasingly move across those products. Asking it to read a syllabus and create Calendar entries might save only a few minutes, but habits are often built around small conveniences repeated until the alternative feels unnecessarily cumbersome.
A year without a subscription charge gives Google time for those conveniences to become familiar. Someone who begins using Gemini during first year to organise notes, revise for exams and conduct research may eventually enter the workplace with a fairly settled idea of what an AI assistant is supposed to do and which one they expect to use. That doesn’t make Google’s offer a trick. The student gets a product that may be useful and Google gets an opportunity to establish a relationship early. Both things can be true without pretending the transaction is more innocent or more sinister than it is.
The question becomes more complicated in education because efficiency isn’t always the only objective. Producing an essay faster isn’t necessarily equivalent to learning more from writing it, and immediately resolving confusion isn’t necessarily equivalent to developing the ability to work through it. Students learn partly through the friction of deciding which information matters, discovering that their first interpretation was wrong, finding better evidence and recognising the limits of what they understand. An AI study system can support those processes, but it can also perform enough of them that the student has to think carefully about what remains theirs.
That problem isn’t unique to Google’s approach. HelloAida, a South African AI tutoring platform, has tried to design its tutoring around guided reasoning rather than simply handing learners answers. Its challenge illustrates another part of the South African AI education problem: pedagogical design means little if the learners who could benefit from it can’t afford sustained access. Google’s student offer removes one price barrier for a year, although access to the hardware and connectivity around Gemini remains uneven.
This gap shows up most clearly when products conceived in much wealthier markets are presented as universally available educational tools. South Africa has relatively high internet penetration while its use of generative AI remains substantially lower, which suggests availability alone isn’t producing equal participation. Students who already have reliable connectivity and suitable devices are likely to extract considerably more value from Google’s free year than students whose access is intermittent, expensive or mostly mobile.
Google isn’t responsible for fixing South Africa’s education system before it can offer Gemini here, nor would withholding useful AI tools somehow produce a fairer university system. The interesting part is how quickly the meaning of “using AI for studying” is expanding. Asking Gemini to explain a difficult paragraph is one kind of assistance. Allowing it to identify what you haven’t learnt, determine what you should study next, generate the material and help organise when you study it is a much larger role.
Universities will spend the next few years deciding where assistance ends and the intellectual work they actually want students to develop begins. Google doesn’t have to answer that question before its products become embedded in the process. If students find Gemini useful enough during the free year, some of those boundaries may be shaped by everyday behaviour before institutions ever manage to draw them.