Inside a PhD Student's Research Workflow | Studying Parasocial Relationships with Alvina Lai
Consensus · video · 2026-06-25 Reader 🖥️
Amplifier / Delegator / Apprentice TypologyLLM Reliability FailuresRetrieval & RAGAI Workflow Patterns
Effective AI-assisted literature research, as Alvina Lai practices it with Consensus, is not a single search step but a repeatable staged workflow — natural-language topic search over a peer-reviewed corpus, triage of algorithmically-flagged key papers, source capture into a library, verification of drafted claims against extracted evidence, a second targeted search to establish a literature gap, and (newly) citation-graph-based discovery — that replaces exact-keyword, paper-by-paper verification with faster, terminology-agnostic retrieval and writing-stage claim-checking.
practical · interview / product-demo testimonial, hosted and produced by the tool vendor's own channel — This is a scoped slice of a longer interview; only the research-workflow/tooling portion (roughly the back half of the video) is extracted, per the user's explicit request. The substantive psychology content about parasocial relationships and polarization is deliberately excluded.
Structure
| Part | Role |
|---|---|
| Topic framing & Deep Search | entry point / broad discovery |
| Key-paper triage & reading | relevance/quality filter |
| Source organization (Library / Zotero integration) | capture and retention |
| Claims & Evidence verification | writing-stage claim-checking |
| Gap-mapping search | novelty justification for the proposal |
| Ask-the-paper quick comprehension | ad hoc convenience use |
| Citation Graph discovery | exploratory / tangential-topic finding — newly added, untested by her prior to this demo |
Key terms
- Deep Search
A Consensus search mode she used to start her grant-proposal literature review from a plain-language description of her topic; returned 352 eligible papers with a stated search methodology and quality ranking, rather than requiring exact keywords. - Claims & Evidence
A Consensus panel she consults while writing to check that the claims stated in her manuscript/proposal are backed by the strongest available evidence in the retrieved literature. - Citation Graph
A newly released (≈1 week before this recording) visualization: from a 'seed' paper it surfaces (1) direct citations, (2) papers consistently co-cited alongside it, and (3) papers sharing similar bibliography; additional seeds can be added and the graph regenerated around the growing set. Note: this mechanism was explained on camera by the Consensus interviewer (Daniel), not by Alvina, since she had just said she'd never used it; Alvina herself, after seeing it demoed, spontaneously analogized it to 'Obsidian... but for academia.' - Library (Zotero integration)
Consensus's built-in reference-saving feature; lets her upload/save sources directly inside Consensus instead of using the standalone Zotero browser extension on every PDF. - Ask the paper
A chat-with-document feature; she uploads a paper (including her own drafts) and asks Consensus to explain or summarize it for fast comprehension when she doesn't have time for a full close read. - Key papers
Consensus's algorithmically flagged subset of the most relevant/highest-quality results within a search; she treats these as the papers to read first and as primary citation sources.
Propositions
- explicitConsensus is the only AI tool Alvina relies on for research, because she deliberately avoids general-purpose LLMs like ChatGPT due to hallucination risk in an academic context that requires source accuracy.conf: high
- explicitConsensus's peer-reviewed-only sourcing lets her trust results without independently verifying peer-review status herself, unlike Google Scholar, which surfaces preprints, opinion pieces and books she can't rely on.conf: high
- explicitConsensus's Deep Search mode lets her start literature review from a natural-language description of her topic rather than exact keywords, retrieving sources across fields that use different terminology for the same underlying concept.conf: high
- explicitThe Claims & Evidence feature let her verify, while writing, that the claims in her grant proposal were backed by the strongest available evidence in the literature.conf: high
- explicitA second, more targeted search pass — checking what interventions already exist for reducing political polarization and whether they work long-term — was how she established that her wisdom-based intervention angle was a genuine gap in the literature, justifying the proposal's novelty.conf: high
- explicitHer Consensus-based literature-review-heavy grant proposal took about two weeks, versus about a month using Google Scholar for a comparable master's proposal — though she notes most of the two weeks went to proposing methodology, not the literature review itself.conf: medium
- explicitHer prior Google Scholar workflow required manually verifying each written sentence's claim against search results one query at a time, because Google Scholar's single search bar can't handle synonym/terminology variation the way Consensus's natural-language search can.conf: high
- explicitConsensus's newly added Zotero-integrated library is a meaningful improvement over her previous workflow of manually invoking a browser Zotero extension for every PDF she wanted to save.conf: high
- explicitThe 'ask the paper' feature lets her get a fast summary/explanation of an uploaded paper for quick comprehension (e.g., late at night) without a full close read.conf: high
- implicitCitation Graph is presented as a favorite feature for discovering niche, tangentially related literature via three relation types from a seed paper (direct citations, papers consistently co-cited alongside it, and papers sharing similar bibliography) — but Alvina had never used the feature before this interview, and the mechanism itself is explained on camera by the Consensus interviewer, not by her.conf: medium
- explicitAfter seeing Citation Graph demoed for the first time, Alvina spontaneously reached for a personal-knowledge-management analogy — 'it's like Obsidian but for academia' — to describe it, rather than this framing originating from the Consensus interviewer as marketing copy.conf: high
- implicitTrusting Consensus because its input corpus is peer-reviewed addresses source-quality risk, but does not by itself guarantee that Consensus's own AI-generated synthesis layer (Claims & Evidence extraction, deep-search categorization, ask-the-paper summaries) accurately represents what the underlying papers say — a distinction the interview never draws, even though it explicitly raises hallucination as the reason to avoid ChatGPT.conf: mediumsupportsEating My Own Dog Food: How I Used the Framework to Write the Post About the Framework · p4 — This document's implicit critical proposition (trusting peer-reviewed inputs doesn't guarantee Consensus's own AI-generated synthesis is drift-free) is supported by the established finding elsewhere that AI-generated citations 'look clean but systematically drift in accuracy across drafts, requiring human verification against primary sources.'
Arguments
Consensus substantially reduced the time and effort needed to complete a literature-review-heavy grant proposal compared to her prior Google Scholar-based workflow. strength: low
Explicit premises:
- Google Scholar required exact-keyword, one-paper-at-a-time claim verification and took about a month for a comparable master's proposal.
- Consensus lets her search by natural-language topic description, surfaces quality-ranked 'key papers,' and provides Claims & Evidence for direct verification of drafted claims.
- This PhD grant proposal took about two weeks with Consensus, under a time crunch.
Implicit premises:
- The two proposals (master's vs. PhD) are similar enough in literature-review difficulty for the one-month vs. two-week comparison to isolate the tool's effect rather than her own growing research skill or topic familiarity.
- The volume/difficulty of available literature didn't independently differ enough between the two topics to explain the speed difference on its own.
Inferences:
- Faster, keyword-independent retrieval plus built-in evidence verification → less time spent sourcing and cross-checking claims → shorter overall proposal-writing time.
AI-assisted, peer-review-filtered search tools like Consensus solve the interdisciplinary terminology/equivocation problem that keyword-based literature search cannot. strength: low
Explicit premises:
- Different fields use different terms for the same construct (e.g., 'parasociality' in her field vs. 'trust' in political science).
- Google Scholar's exact-keyword search can't bridge that terminology gap.
- Consensus's natural-language Deep Search retrieved a broad, quality-ranked set of 352 eligible papers from a single topic-description prompt.
Implicit premises:
- Consensus's underlying retrieval performs genuine semantic/conceptual matching across the terminology gap, rather than simple synonym expansion of the literal prompt text.
Inferences:
- A large, high-quality result set returned from one natural-language prompt → therefore Consensus solves cross-field terminology mismatch.
Narratives
- The grant-proposal literature-review walkthrough: she prompted Deep Search with a plain description of her 'wisdom as a buffer against political parasocial relationships' topic, got back 352 eligible papers, triaged Consensus's flagged 'key papers,' read/saved PDFs into a project folder, cross-checked Claims & Evidence, and finished a proposal she says took about two weeks — versus roughly a month via Google Scholar for a comparable earlier proposal.
→ Illustrates the full staged workflow end-to-end and anchors her central time-savings claim. - Live in-interview Citation Graph demo: the interviewer asks her to pick a 'niche' paper (she chooses one on 'conspiracy ideation'), hits 'seed and regraph,' and narrates the three link types while she operates the tool on screen — after she had already admitted minutes earlier she'd never tried the feature. Watching it work, she then spontaneously calls it 'like Obsidian but for academia.'
→ Demonstrates a feature in real time, but the surrounding admission reveals the endorsement is a first impression rather than tested practice; her unprompted Obsidian/PKM analogy is a more organic signal of perceived value than the vendor's own narration of the mechanism. - The 'ask the paper' 3 a.m. anecdote: while working late on a proposal, she uploads her own paper and asks Consensus to explain it because she's too tired to do a close read before falling asleep.
→ Illustrates a specific, low-stakes convenience use case distinct from her higher-stakes literature-review and claim-verification uses.
Questions addressed
- ✅ How can a researcher search literature across disciplines that use different terminology for the same concept?
Use natural-language topic prompts via Consensus's Deep Search instead of exact-keyword search — though the mechanism by which this bridges terminology gaps is not explained or demonstrated. - ✅ How does she identify a genuine literature gap to justify a grant proposal's novelty?
Run a second, targeted search on existing/adjacent interventions, check whether they work long-term, and combine self-found papers with papers supplied by her PI. - ✅ How does she verify that her own written claims are actually supported by the literature?
Claims & Evidence panel review during the writing stage, replacing her old sentence-by-sentence Google Scholar verification. - ✅ How does she organize and retain sources across a project?
Consensus's new Zotero-integrated library replaces a manual per-PDF Zotero-extension workflow. - ⭕ Does Citation Graph actually improve literature discovery for her research?
Left genuinely open — she flags it herself as untested ('I've no idea what it does yet') immediately before demoing it live for the first time. - ⭕ What are Consensus's own limitations, costs, or failure modes as a research tool?
Gaps
- incompleteThis is promotional content produced by Consensus's own channel, hosted by a Consensus team member, interviewing an enthusiastic user. Every critical comparison in the video targets Google Scholar or general-purpose LLMs (ChatGPT) — no negative or critical account of Consensus's own limitations (cost, coverage gaps, possible errors in its AI-generated Claims & Evidence or search categorization) appears anywhere.
Read next: Treat all Consensus-specific efficacy and accuracy claims in this video as untested self-report from a satisfied user in a testimonial/marketing context, not as independently verified evaluation. - illogicalThe interview conflates 'benefits specific to Consensus' with 'benefits of AI-assisted research generally.' Alvina's hallucination-avoidance reasoning is applied to justify avoiding ChatGPT, and her trust in Consensus rests on its peer-reviewed input corpus — but the same scrutiny is never turned on Consensus's own AI-generated synthesis layer (Claims & Evidence extraction, deep-search categorization, ask-the-paper summaries), which could misrepresent even accurate, peer-reviewed sources despite the inputs being trustworthy.
Read next: When evaluating any AI research tool (Consensus or otherwise), separate 'is the input corpus reliable' from 'is the tool's AI-generated synthesis of that corpus reliable' — the video answers only the former. - incompleteThe video presents established-practice claims (Deep Search, Claims & Evidence, Zotero library) alongside first-use enthusiasm (Citation Graph, demoed live minutes after she admits never having tried it, with its mechanism narrated by the Consensus interviewer rather than described from her own usage) with the same confident framing, without distinguishing tested workflow value from a fresh, untested impression.
Read next: Weight the Deep Search / Claims & Evidence / library claims (repeated practice, specific anecdote, quantified time comparison) more heavily than the Citation Graph endorsement (single live first-use, no track record). - uninformedNo discussion anywhere of Consensus's cost, access model, or institutional licensing — nor whether her University of Toronto affiliation subsidizes or otherwise enables her use of it.
Read next: Readers evaluating whether to adopt this workflow themselves should independently check pricing/access, since the video gives no information on transferability outside an apparently well-resourced academic setting.