On this page
Ask a chatbot how many questions you must attempt in JEE Main, or which chapters are in this year's NEET-UG syllabus, and you will get a confident answer. It may be last year's answer. Exam rules can change from one year to the next, and most AI models learn from text that stops at a fixed date.
What is an AI training cutoff?
A large language model learns from a huge collection of text gathered up to a certain date, and it learns nothing after that unless a tool fetches new pages for it. Anthropic's models overview lists two dates for each model: a reliable knowledge cutoff, which it defines as "the date through which the model's knowledge is most extensive and reliable", and a training data cutoff, which covers the broader range of data used.
Even the published date is only a rough guide. A 2024 study led by Jeffrey Cheng, Dated Data: Tracing Knowledge Cutoffs in Large Language Models, found that "effective cutoffs often differ from reported cutoffs". One reason it gave is that new web crawls still contain non-trivial amounts of old pages. So a model may know less about recent months than its stated cutoff suggests. Our explainer covers how large language models work.
Why would AI give me the old exam pattern?
Take a real JEE Main change. In its public notice of 17 October 2024, NTA said that since JEE (Main) 2021, candidates could attempt any 5 of 10 questions in Section B of each subject. This was a temporary measure brought in because of COVID-19. From 2025, Section B has only 5 questions per subject, and you must attempt all 5.
That rule ran for four exam years, so many older web pages still describe the option to choose. A model trained on that text can repeat the old rule with confidence.
The second example is NEET-UG. NTA's public notice of 8 January 2026 says the NEET (UG)-2026 syllabus was finalised by the Under Graduate Medical Education Board of the National Medical Commission, with an NMC notice dated 22 December 2025. It adds that NTA will set the exam "strictly based on the syllabus as finalized and notified by NMC". Any model whose training data stops before late December 2025 cannot have learned that syllabus from its training data.
Web search helps, but not fully. A chatbot that searches can still land on an unofficial blog or a page about last year's exam. It can also mix up details from different sources, which is the same habit behind AI making up sources and citations.
Where can I check the official syllabus and exam pattern?
Go to the body that runs the exam or sets the syllabus:
JEE Main: NTA's JEE Main website and nta.ac.in. NTA's own October 2024 notice told candidates to keep visiting these two sites for the latest updates.
NEET-UG: the NEET website for notices and the information bulletin, and the National Medical Commission for the syllabus itself. Our guide to the NEET UG syllabus from NMC covers what is in it.
CBSE board exams: the CBSE academic website, which posts a separate curriculum for each academic year. Its menu lists curricula from 2023-24 to 2026-27, so check that the year matches yours.
A five-minute check before you trust an AI answer about your exam
Ask the AI which year and which document its answer is based on. If it cannot name an official notice or information bulletin, treat the answer as a guess.
Open the official site yourself. Check where any chatbot link really leads.
Find the newest notice or bulletin for your exam year, and look at its date.
Compare it line by line: number of questions, sections, choices, marking, and the list of units.
Write down the notice name and date on your planner.
Here is a prompt you can copy: "Tell me what the JEE Main exam pattern was in the last year you have reliable information about, and name the official NTA document. Then list what I should confirm on jeemain.nta.nic.in for this year." A careful model will tell you what it does not know. If it sounds sure about this year without naming a source, be wary.
Parents can use the same routine. If a WhatsApp forward or a chatbot claims a chapter has been dropped, ask for the notice. Official notices, like the two above, carry a date and the issuing body's name.
So what is AI good for in exam prep?
Physics, chemistry and biology concepts do not change when the pattern does. Once you have the official syllabus, AI is useful for learning what is on it. Fermi's AI tutor works out what you already know and builds a plan for the session, so you can bring the units you confirmed on the official site and start with a clear next step. For a single topic you are unsure about, the Fermi Concept Library gives each concept its own lesson.
Bookmark the official site for your exam today, and check it before you trust anyone's summary, including ours.
