This is the second in a series of articles exploring the rise of AI-powered homework helpers. Read part one here.
AI companies are rapidly becoming major players in the education sector, investing heavily in their own generative AI tools designed to support student learning. To put these tools to the test, I used standardized test questions from sources like the New York Regents Exam, New York State Common Core Standards, AP exams from 2024, and the Southern Poverty Law Center’s (SPLC) Learning for Justice program. My goal was to evaluate these STEM-focused bots in subjects aligned with my expertise, while also simulating how an “average” student might use them.
Following a stint with ChatGPT, my second round of AI tutor tests involved Gemini’s Guided Learning — Google unveiled free Google AI Pro plans, along with the new learning mode, to all college students back in August. I utilized a Gemini 2.5 Pro account, ensuring it was set to Guided Learning (accessible by clicking the three dots to toggle this setting).
I posed the same standardized exam questions — using the same initial prompts — to Gemini, ChatGPT, and Claude. I kept my approach simple, using prompts like, “I need help with a homework problem,” and “Can you help me study for an English test?” I provided minimal additional information about my student persona, including grade level, and covered several subjects:
- Math: An Algebra II polynomial long division question from the New York State Regents Exam
- Science: An ecology free response on the impact of invasive species from the 2024 AP Biology test
- English Language Arts: A practice analysis of Ted Chiang’s “The Great Silence” from the New York State Regents Exam
- Art History: A short essay on Faith Ringgold’s Tar Beach #2 from the 2024 Art History test
- American History and Politics: An essay prompt on how American housing laws exacerbated racial segregation, taken from the SPLC’s Learning for Justice program
Here’s my assessment of Gemini as a virtual teacher.
Gemini: The T.A. Who Really Loves Quizzes
Math: A Visual Learning Boost
Gemini emerged as my preferred math tutor. Like ChatGPT, it provided answers concisely, but went a step further by offering a visualization element. As I relearned polynomial long division, Gemini utilized its coding box to approximate the standard long division formatting with small dashes, forming the familiar sideways “L” shape. This made it easy to follow the steps of a subject I’s had long forgotten, appealing to my need for visual aids. Gemini was the most structured and clear math teacher, stopping me when I achieved the correct answer, explaining how to write it on my exam, and adding details needed to achieve full credit (based on the problem I shared).
“Gemini will plot things for you, it kind of writes it like a human would write,” noted Hamsa Bastani, associate professor at the University of Pennsylvania’s Wharton School and co-author of the study “Generative AI Can Harm Learning.” She suggested that GPT-5 might be better at solving math problems through brute force comparison, but most would likely agree that Gemini excels at writing—and ironically, this strength makes it surprisingly adept at explaining math concepts.
Biology Blunder and Reading Comprehension Struggles
Gemini immediately faltered during my AP Biology test. Unlike other chatbots, it didn’s ask personal questions regarding my preferred study methods or the specifics of my test. Instead, it promptly generated a randomized, multiple-choice biology exam covering a range of subjects, prompting me to create flashcards on topics I missed. It only offered free response options when directly asked. These responses were written according to Gemini’s syllabus.
The bot’s enthusiasm for quizzes resurfaced during the English Language Arts question. Responding to “Can you help me study for an English test?” Gemini explained, “I can do a lot of things to better your studying; what do you need help with specifically?” I replied that my fictitious teacher, “The College Board,” had assigned a practice test. Gemini generated short passages—chatbot-written approximations of famous works—but with a distinctive writing style. The first, titled “The Road Not Taken,” bore a slight resemblance to Robert Frost’s poem, yet the writing clearly wasn’s his. “We stand today at a crossroads. Down one path lies the comfortable and familiar, the road of complacency,” it stated. This wasn’t how I remembered the poem—was this what a chatbot thought “two paths diverge in a yellow wood” meant?
The inability to pull complete copies of existing texts, a common issue with AI chatbots, likely stems from ongoing copyright concerns. Anthropic recently settled a $1.5 billion class action lawsuit filed by authors whose works were used to train its AI. While unable to retrieve full texts, Gemini was the only chatbot to offer these approximations of classic literature unprompted.
Despite this drawback, Gemini’s user experience included a significant advantage: it displayed its reasoning step-by-step, accessible by clicking the “Show thinking” drop-down menu at the top of each response. This proved helpful for understanding why Gemini addressed certain portions of my prompts and how it reasoned through my incorrect answers.
Social Science Surprise: Essay Assistance
While failing to engage effectively for reading comprehension lessons, Gemini surprisingly excelled at drafting social science essays and short answers. For Art History, it broke down my responses constructively without being overly critical or rewriting my responses. When I requested help with an essay on housing discrimination, it happily encouraged me to take the lead, asking me to explain my existing knowledge and organize it into a simple essay structure. It left blanks for me to fill in with information from my own lessons, avoiding the writing of any text itself (as I hadn’s requested it).
Bastani wasn’s surprised by this discrepancy: “It’s very good at some tasks, and then it’s not great at other tasks that are very similar-looking. And you have to be an expert yourself to be able to recognize the difference.” Ethan Mollick, Bastani’s colleague and author of Co-Intelligence: Living and Working with AI, calls this AI’s “jagged frontier”—an invisible wall delineating related tasks an artificial intelligence can and cannot logically complete. Tasks appearing close may actually be on opposite sides of the wall, and users may be unaware.
Therefore: literature analysis falls outside the wall; an essay about racial segregation is inside the wall.
Summing it Up
Gemini Guided Learning Pros:
- My preferred math tutor, offering a unique visualization element.
- Offers various learner options, including flashcards, quizzes, and study guides.
- Accessible and straightforward voice.
Gemini Guided Learning Cons:
- Struggles with reading comprehension.
- Quick to serve users unhelpful, automatically-generated quizzes and flashcards.
- Like its competitor, ChatGPT, it emphasizes rote practice as key to learning
