Thuta Learning
AdvancedProductivitybeginner

AI-Assisted Research

What you'll walk away with

  • Explain the core ideas behind AI-Assisted Research
  • Read the diagram and trace how information or tasks flow through the workflow
  • Explain how this applies to your own personal system

Build the mental model

AI-assisted research follows the same pipeline as ordinary research, folding AI into just two steps: angle generation and comparison. AI never replaces search, evaluation, or synthesis.

1. Question

Define a real research question.

2. AI generates search angles

Ask an AI tool for plausible angles; this is a search plan, not findings.

3. Search real sources

Search for real sources against each angle.

4. Read and take notes

Read every source yourself and record a structured note.

5. AI helps compare

Use AI to compare notes and surface agreement or contradiction patterns.

6. Human verifies

Manually verify every citation and claim — this step cannot be skipped.

7. Output

Produce the output only after verification.

text
AI-ASSISTED RESEARCH FLOW
-------------------------
AI-ASSISTED RESEARCH FLOW
--------------------------
[question] -> [AI: search angles] -> [search real sources]
                                              |
                                              v
                                     [read + take notes]
                                              |
                                              v
                                   [AI: compare notes]
                                              |
                                              v
                              ==[ HUMAN VERIFICATION ]==
                                              |
                                              v
                                          [output]

Connect it to a real scenario

Start by asking an AI tool for five or six distinct angles on your question, not the answer itself. Feed each angle into a real search.

Read every source yourself before it enters your notes. Do not trust a single AI summary — it can quietly drop the caveat that changes your conclusion.

  • Once you have structured notes, ask AI to compare the claims
  • Treat the output as a draft comparison, not a final judgment
  • Open and verify every citation by hand

If you skip verification

Even a citation with a real-looking DOI or URL can be fabricated if you trust it without opening it — that is how invented sources end up in real output.

Never trust AI-generated citations blindly

An AI tool can produce a citation that looks perfectly formatted and still be entirely fabricated. Before anything reaches your output, open every citation by hand and confirm it exists and actually supports the claim.

Try the working example

javascript
function flagCitations(citations) {
  const doiPattern = /^10\.\d{4,9}\/\S+$/;
  const urlPattern = /^https?:\/\/[^\s]+\.[a-z]{2,}(\/[^\s]*)?$/i;
  const placeholderWords = ["example", "citation needed", "source name", "journal of somewhere", "n.d."];

  return citations.map((c) => {
    const text = c.trim();
    const lower = text.toLowerCase();
    const looksLikeDoi = doiPattern.test(text);
    const looksLikeUrl = urlPattern.test(text);
    const looksFabricated = placeholderWords.some((w) => lower.includes(w));

    let status;
    if (looksFabricated) status = "needs_verification"; // placeholder-shaped, likely invented
    else if (looksLikeDoi || looksLikeUrl) status = "check_it_resolves"; // real shape, still confirm it loads and says this
    else status = "needs_verification"; // no verifiable identifier at all

    return { citation: text, status };
  });
}

const aiCitations = [
  "10.1145/3442188.3445922",
  "https://arxiv.org/abs/2005.11401",
  "Example et al. (2024), Journal of Somewhere, p. 42",
  "Smith, J. (n.d.). Citation needed for this claim.",
  "https://openai.com/research/gpt-4",
];

for (const r of flagCitations(aiCitations)) {
  console.log(r.status.padEnd(20), "-", r.citation);
}
You should see
check_it_resolves    - 10.1145/3442188.3445922
check_it_resolves    - https://arxiv.org/abs/2005.11401
needs_verification   - Example et al. (2024), Journal of Somewhere, p. 42
needs_verification   - Smith, J. (n.d.). Citation needed for this claim.
check_it_resolves    - https://openai.com/research/gpt-4

Even "check_it_resolves" does not mean fully verified — it only means the shape is plausible; it still has to be opened and read.

5-minute try-it

Ask an AI tool for five citations on a research question of your own. Run them through flagCitations(), then actually open every "check_it_resolves" citation and confirm it supports the claim it is attached to.

One important caution

Treating an AI's angle list as findings and writing without actually searching

Assuming a citation is verified just because its format looks correct

Wikipedia: Hallucination (artificial intelligence)Productivity Systems

Easy traps

  • Treating an AI's angle list as findings and writing without actually searching
  • Assuming a citation is verified just because its format looks correct
  • Switching productivity systems every week is often a sign the system itself isn't the real problem -- pick one and give it a few weeks before judging it.

Exercise

Ask an AI tool for five citations on a research question of your own. Run them through flagCitations(), then actually open every "check_it_resolves" citation and confirm it supports the claim it is attached to.

You'll know it worked when: check_it_resolves - 10.1145/3442188.3445922 check_it_resolves - https://arxiv.org/abs/2005.11401 needs_verification - Example et al. (2024), Journal of Somewhere, p. 42 needs_verification - Smith, J. (n.d.). Citation needed for this claim. check_it_resolves - https://openai.com/research/gpt-4 Even "check_it_resolves" does not mean fully verified — it only means the shape is plausible; it still has to be opened and read.

AI-Assisted Research | Thuta Learning