profile

The Effortless Academic

Your Complete Guide to Consensus AI for Literature Reviews

Published 5 months ago • 2 min read

In Depth Review of Consensus AI for Lit Reviews

Dear Scholar,

Every week, I am approached by new AI companies that offer help with academic writing or literature searches. It is easy to want to jump to the next thing and proclaim, "AI just killed lit reviews". This is why I decided to go instead in-depth and drill into every detail of the Consensus App.

This tool has been around for years and was routinely featured in the ChatGPT store. I have introduced it as part of the Effortless Literature Review Webinars and use it frequently in my own work. If you want to know every single detail about how this tool works, click the link below.

In other news, ChatGPT released a new update this week: GPT4-omni. It is an impressive advancement over ChatGPT-4, being faster and more "emotionally intelligent" – some information on the release and what it means for academics is below.

​Review: Consensus AI – An Intelligent Literature Review Companion (20 min read)

Consensus features essentially three main features. (1) Literature search using semantic AI to find literature from plain text queries. (2) Automated extraction of insights, especially for yes/no questions (e.g. Consensus Meter in the screenshot). (3) Processing insights from papers into any format you need using the Copilot (e.g. a lit review outline). This article explains the limitations of Consensus and what you should and should not use it for. If you are interested, a 30% discount coupon is available. This is an in-depth article with 17 infographics and screenshots.

Thew new GPT4-omni update:

It can talk in real time, capture clues like laughter, intonation and even video input, thanks to its new architecture.

How it works

Talking to ChatGPT is not new, but it is done differently now. Previously, there were three models:

  • One translated audio to text
  • then the normal ChatGPT answered your query
  • and then another model translated that text to audio.

This was slow and, most importantly, missed key communication cues like voice, intonation and so on.

In the new model, there is a single model that is not just text→ text but directly audio→ audio. It does not require models 1 and 3 anymore. The key benefit is it's faster, can tell multiple voices apart and feels much more natural.

What does it mean for academics?

Check out my thread if you want to see demo videos and a comparison with the previous GPT4 model,

Generally, the model is better at understanding images like figures and can be used as a real-time tutor for learning concepts like math. While adaptation will take time, the technology is promising in helping learn new topics and acts as a fallback for explaining confusing scientific papers. At this point I can only see it as an (increasingly powerful) assistant rather than a replacement to the researcher.

Ilya Shabanov,

The Effortless Academic​

Don't want to receive emails like this: Unsubscribe from newsletter only .
Want to unsubscribe from all future emails? Unsubscribe from everything.
Your Preferences.

The Effortless Academic

Literature Review Tools, Note-Taking Strategies and AI tutorials for the modern academic. Publish more with less effort and supercharge your career.

Read more from The Effortless Academic

Deep Dive: Nested Knowledge for Systematic Literature Reviews Dear Scholar, Systematic Literature Reviews require a lot of organization, patience, and tedious database screening. AI search engines could not help here, as they do not meet the rigorous requirements of a systematic literature review. Nested Knowledge saw this problem and found a solution that automates tedium and uses AI smart This tool covers all the needs of systematic reviews: systematically finding papers, downloading them,...

4 days ago • 1 min read

Don't start your lit review with a question Dear Scholar, Whenever you look for advice on literature reviews, it always starts with defining either a question or a topic. But this is backwards since good questions can only come once you are substantially familiar with something. Instead, I suggest starting your literature review with an overview, drilling down to your niche, learning it intimately and then, as your last step, defining the question. This article explores one of the many topics...

11 days ago • 2 min read

Research Maps are the key to making new discoveries. Dear Scholar, The most frequent question people ask me during my literature review webinars is, "Have you tried tool X"? The answer is yes; I have tried almost every tool, but most of them are not mature enough and have not stood the test of time. While the tool is not as important as your ability to use it and your knowledge, a good tool can greatly speed up your literature review. This is what this week's email is about. In other news, I...

17 days ago • 2 min read