Report on CHIIR 2026 Workshop on Generative AI and Academic Search (GAI&AS)
This report summarizes the CHIIR 2026 GAI&AS workshop, which brought together researchers to explore how generative AI is transforming academic search systems beyond traditional retrieval to support synthesis and learning while emphasizing transparency, credibility, and human-centered design principles.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the world of academic research as a massive, ancient library. For decades, the librarians (the search engines) had one job: find the right books on the shelf and hand them to you. You, the researcher, were responsible for reading them, connecting the dots, and figuring out what they meant.
Now, a new kind of librarian has arrived: Generative AI. This librarian doesn't just fetch books; it reads them for you, writes summaries, answers your questions, and even tries to have a conversation about the material.
This paper is a report on a meeting of experts (the CHIIR 2026 Workshop) who gathered to ask a big question: Is this new librarian helping us learn, or is it doing the thinking for us?
Here is a breakdown of what they discussed, using simple analogies:
1. The Core Problem: The "Magic Box" vs. The "Learning Journey"
The experts noticed a shift. In the past, searching for information was a journey where you had to do the heavy lifting of understanding. Now, the AI gives you the "answer" immediately.
- The Analogy: Imagine a student trying to learn how to bake a cake.
- Old Way: They read the recipe, measure the ingredients, and mix the batter. They learn how to bake.
- New Way: They ask a robot to bake the cake for them and hand them the finished product.
- The Concern: If the robot does all the work, the student might get a cake, but they never learn the skill of baking. The workshop worried that in academia, students and researchers might stop doing the critical thinking required to understand their own fields.
2. The Three Main Themes of Discussion
The group organized their thoughts into three main buckets:
A. The Foundation (Building a Trustworthy Librarian)
Before we let the AI do more, we need to make sure it's honest and transparent.
- The "Black Box" Problem: Currently, AI often gives an answer without showing its work. It's like a magician pulling a rabbit out of a hat without showing you how.
- The Goal: The experts want AI to show its "receipts." It needs to say, "I found this answer because I read these three specific papers," so researchers can check if the AI is lying or making things up (hallucinating).
- Bias Check: They found that AI tends to favor famous researchers and ignore lesser-known ones, much like a news channel that only interviews celebrities and ignores local experts. This creates a "rich get richer" problem in science.
B. The Applications (What Can the Librarian Actually Do?)
The group brainstormed how to use this new tool without losing control.
- The "Co-Pilot" Idea: Instead of the AI driving the car, it should be the GPS. It can suggest routes (search queries) or summarize traffic (summarize papers), but the human should still hold the steering wheel.
- New Jobs for AI: They imagined AI helping with specific tasks like:
- Organizing a messy pile of research papers.
- Finding gaps in current knowledge (where no one has looked yet).
- Translating complex jargon into plain English for non-experts.
- The Challenge: We need to figure out exactly where the line is between "helpful assistance" and "doing the work for you."
C. Search-as-Learning (The "Friction" Factor)
This was a major point of discussion. Sometimes, the struggle to find an answer is actually good for learning.
- The "Friction" Analogy: Think of learning to ride a bike. The wobbly, difficult part is where your brain builds the muscle memory. If a robot held the bike perfectly steady for you, you'd never learn to balance.
- The Insight: The experts argued that AI shouldn't remove all the "friction" (the difficulty) of research. We need to design systems that force users to stop and think, rather than just accepting the first answer the AI gives. They want to build "scaffolding" (like training wheels) that helps you learn, but eventually comes off so you can ride on your own.
3. The Big Takeaways
The workshop concluded with a few clear messages for the future:
- Don't just chase speed: Just because AI is fast doesn't mean it's better. We need to value the quality of the thinking process, not just the speed of the answer.
- Human Agency is Key: The human must remain the boss. The AI is a tool, not a replacement for the researcher's brain.
- We Need New Rules: The old ways of studying how people search for information don't work anymore. We need new theories and methods to understand how humans and AI work together.
- Collaboration: Librarians, teachers, computer scientists, and researchers need to work together to build these tools. It's not just a tech problem; it's a human problem.
Summary
In short, this paper is a call to action. It says: "We have a powerful new tool that can change how we do research, but if we aren't careful, it might make us lazy thinkers. We need to design these tools to help us learn and think critically, not just to give us easy answers."
The experts plan to keep talking about this, inviting more people (like librarians and industry developers) to join the conversation to ensure the future of academic search supports human intelligence rather than replacing it.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.