How to find scientific papers with artificial intelligence

Before opening anything, write in one line what you want to know. The more specific the question, the better the studies the AI brings you.

• Vague: “exercise and health”.
• Specific: “Does strength training reduce anxiety in young adults?”

Good questions usually have a subject (young adults), an action (strength training) and an outcome (anxiety). If in doubt, ask an AI like Claude or ChatGPT to help you sharpen it: “Turn this idea into a clear research question”.

Consensus is ideal to start: you type your question as you’d ask a person and it answers based on real papers, showing whether the evidence says yes, no or it’s split.

1. Go to https://consensus.app and create an account (it has a free tier).
2. Type your question in the search bar and press Enter.
3. Read the summary at the top (the “consensus”) and scroll to see each study behind it.
4. Tap a study’s title to open the original paper.

Start here when you want a quick answer and to see at a glance where the evidence points.

When you already have your question and want to compare several studies, Elicit shines: it builds a table where each row is a paper and each column a data point (sample, method, result).

1. Go to https://elicit.com and create your account (free tier with a monthly limit).
2. Type your research question.
3. Elicit shows you a list of studies with their summary.
4. Add columns like “number of participants” or “main finding” and compare at a glance.

It’s like having an assistant that reads the papers and fills the table for you. You decide which ones to open in depth.

Semantic Scholar is a free search engine with over 200 million papers. It’s perfect to dig deeper and find related studies.

1. Go to https://www.semanticscholar.org (no account needed to search).
2. Type the topic or paste the title of a study you already liked.
3. Open a paper and check the “Citations” and “References” sections to jump to connected studies.
4. Use the filters (year, field, open access) to keep only the most relevant.

Trick: when you find ONE good study, use its citations and references to discover the whole conversation around the topic.

Finding the paper is half of it; the other half is understanding it. Here an AI like Claude or ChatGPT helps a lot, especially if the study is in English.

1. Open the PDF or copy the paper’s abstract.
2. Paste it into Claude and ask: “Summarize this study in simple terms: what they researched, how, and what they found”.
3. Ask anything you don’t get: “What does this term mean?” or “What limitations does this study have?”.

If you want to go deeper on getting the most out of this, read our tutorial on how to use Claude for text analysis and the one on how to summarize long texts.

AI speeds up the search, but the responsibility is still yours. Before using a study:

• Open the original paper, don’t rely only on the summary the AI generated.
• Confirm the study says what you think (sometimes the nuance changes everything).
• Check the date and journal: prefer recent studies from serious sources.
• Cite the real source yourself, with its author and year.

Think of AI as a lightning-fast librarian: it brings you the right books in seconds, but reading them and citing them well is up to you.