AI study tools: which kind do you actually need?

Search for AI study tools and you get a list of forty apps. They are not forty versions of the same thing. They fall into about five kinds, and each kind solves a different problem, so the first question is not "which app" but "which problem". This guide names the kinds and the main products in each, says what each is good and bad at, gives a decision table, lists what to check before trusting any of them with your material, and says honestly where Bord sits.

The five kinds, with examples

Match the tool to the problem

Name the problem first, then pick the kind. Most students have two or three of these problems, and a tool from each kind is normal.

If your problem is understanding, which for most courses is the actual problem, you need to see the parts of a topic at once, test yourself on them, and keep the result somewhere you will open again. That is what a board is for, and it is the one row where there is not much choice yet.

Chatbots

ChatGPT and its relatives are the tools most students already have. They are patient, fast and free at the level most studying needs. They are best used for one-answer questions: a definition, a step you do not follow, a sanity check on a calculation.

Where they fail is shape. Answers scroll away, comparisons come one above the other, your own material gets pasted into the same scroll as everything else, and the output is read-only. The research is also a caution: a large controlled study found students who studied with an unrestricted chatbot scored worse on the exam than students with no AI, because they asked for answers and copied them [1]. There is a full treatment in ChatGPT for studying. Short version: use it for quick questions, make it quiz you rather than tell you, and move anything worth keeping into real notes.

Flashcard apps

Anki is the standard: free on desktop, open source, with a spaced-repetition scheduler that shows each card just before you would forget it. Quizlet is the friendlier commercial version with a huge library of shared sets. Knowt and RemNote add AI that generates cards from your notes or slides, which solves the main reason decks die: nobody wants to write five hundred cards.

Spaced repetition is one of the two study techniques with the strongest evidence behind it [2], and for facts there is nothing better. The limit is that flashcards test recognition of discrete items. They are poor at "explain why", at diagrams, and at anything where the relationship between parts is the point. Medicine and language students should have one. Students in subjects built on mechanisms and arguments should treat it as a supplement.

Source notebooks

NotebookLM, from Google, is the best known. You upload PDFs, slides, web pages or recordings, and ask questions; answers come back with citations to the source, and it can generate a study guide or an audio overview. Mindgrasp, YouLearn and others do similar things with different extras.

These are the right tool when the problem is volume: a long reading list, a textbook you need to search rather than read. The citations are a real advantage over a chatbot, because you can check. The limit is that the output is still text in a panel. You read the answer, and reading an answer is the low-utility end of studying [3]. The notes you make from it live somewhere else.

Lecture recorders

Otter is a general meeting transcriber that many students use for lectures. Turbo AI, Notewise and a few others are built for students: record the lecture, get a transcript, a summary, sometimes flashcards and a quiz. Bord's Live Notes is in this kind too, with the difference that the notes are built onto a board as the lecture goes rather than produced as a document afterwards.

The value is obvious: you listen instead of transcribing, and nothing is missed. The trap is also obvious: a recording you never replay and a summary you never rework are not notes. Whatever recorder you use, the twenty minutes after the lecture, tidying and rewriting in your own words, is where the method pays off. There is a comparison of recording with handwritten and typed notes in The best way to take lecture notes.

AI whiteboards

The newest kind, and the one with the fewest products. The AI writes and draws on an infinite canvas, as objects you can edit, and reads what is on the board when you ask the next question. A comparison becomes two columns; a process becomes a chain; a diagram comes out as labelled parts rather than a locked picture. The full definition, and a checklist for telling a real one from a drawing app with a chat panel, is in What is an AI whiteboard?

It fits the problem the other kinds do not solve: understanding a topic with structure, and keeping the result as notes you will revise from. The evidence for laying ideas out visually is good, with a meta-analysis finding concept maps beat text and outlines for retention, most of all when the learner edits the map [4]. The limit is that a board is a poor place to read long prose, and its flash cards sit beside your notes rather than in an app built only for review.

What to check in any of them

Before you put a term's worth of material into any tool, check these. The answers are usually in the privacy policy and the pricing page, and a tool that hides them is telling you something.

• Are your files and notes used to train models? Bord's are not. Several free tools' are, by default.
• Do answers from your material cite the page or the timestamp? If not, you cannot check them without rereading.
• Can you try it without an account? Bord gives guests free weekly usage with no sign-in. Adding files needs an account in nearly every tool, because a file is an upload.
• Can you fix a wrong note yourself, or do you have to re-ask? On a board you edit it. In a chat you re-ask.
• Can you export your work? Notes locked inside an app are notes you lose when the subscription ends.
• What happens to a deleted file? Bord deletes the file, its figures and its index. Some tools keep derived data.

Where Bord sits

Bord is an infinite notebook and whiteboard with the AI on the board. It covers all five kinds above from one place. Ask explains on the board, with web search when a fact needs to be current. Drop in PDFs, slides, photos, audio and video as sources, and every answer cites the page or the moment, which is the source-notebook job. Live Notes records the lecture and builds the notes while you listen, which is the recorder job. Quizbook tests you with multiple choice, gaps, written answers graded on reasoning, and diagrams you label or draw, then gives you a report card. Flashcards writes a deck from a topic or your files; you flip and mark each card, and it comes back when it is due, which is the flashcard-app job. Tutor teaches out loud and draws as it speaks, which none of the other kinds do.

What it does not do is essay drafting, and its decks live on the board rather than in a review-only app. If you have a thousand drug names to drill over years, Anki still earns its place beside it. For the terms, steps and formulas of a course, the deck on the board is enough.

Questions