Whether artificial intelligence is seen as friend or foe, it is undoubtedly transforming various aspects of our work, including how we discover content, information, and knowledge. In our initial post on AI and information discovery, we discussed the evolution of AI and its potential to address pain points for researchers and publishers alike.
Here, we delve into generative AI (GenAI). These systems can produce original and realistic outputs based on the patterns and data they have been trained on. We’ll discuss how GenAI is advancing us towards conversational discovery and what this might mean for publishing, as well as potential future trends in information discovery.
The Evolution of Information Discovery with GenAI
We are witnessing a technological evolution in information discovery applications, along with a shift in how researchers use these tools. Public search engines serve as a compass in a vast sea of information, while commercial and researcher-facing discovery engines focus on open-ended reasoning, simplifying literature reviews, and saving time. Bibliography databases act as massive libraries holding metadata and information about millions of books, research papers, and articles, helping researchers uncover valuable insights and trends. Let's explore these applications in more detail.
GenAI Native Applications
Most of us have likely used ChatGPT to quickly find answers, and the same applies to researchers. ChatGPT's ability to analyze large amounts of information and data saves time and effort. However, it has limited updated knowledge, with ChatGPT 3.5 providing information up to April 2023. ChatGPT 4, Google Gemini (formerly Bard), and Microsoft Bing Copilot can search the internet for up-to-date information, streamline the process of identifying relevant scholarly content, create and analyze images and datasets, and support contextual conversations and AI-augmented answers.
Microsoft aims to leverage its advances in GenAI (relying on OpenAI technology) to boost Bing's usage, which currently holds 3.6% of the global search engine market. In contrast, Google seeks to enhance its GenAI usage and compete with ChatGPT, leveraging its dominance in the search engine market (90.9%). Perplexity.ai, a new conversational search engine, offers a clean and simple way to discover information, focusing on scholarly content.
AI-Powered Bibliography Databases
Familiar scholarly databases like Scopus and Dimensions have integrated GenAI into their platforms. AI-powered features include natural language search, concise summaries, and research synthesis based on vast databases of trusted content. Scopus AI uses Elsevier’s LLM technologies along with GPT from OpenAI for ChatGPT, generating concept maps from research abstracts. Dimensions Assistant provides well-structured explanations, searches licensed literature securely, and notifies researchers of new content based on their queries.
Commercial and Research-Facing Discovery Tools
Tools like Elicit, Scite, SciSpace, and Consensus enhance the research process:
Elicit: Offers customized data extraction and a 'List of Concepts' feature summarizing main research topics.
Scite: Specializes in reference discovery and identifying supporting or contradictory references with three alert mechanisms.
SciSpace: Features a Copilot assistant explaining papers and answering user queries.
Consensus: Leverages GPT-4 and other LLMs to summarize research results, with many tools connected to ChatGPT as GPT plug-ins.
Commonalities of GenAI-Powered Search Tools
Returning Answers, Not Just Results
GenAI enables natural language search, enhancing content discoverability by providing exact answers to queries, even when users lack precise terminology. This approach allows for multilingual queries, increasing accessibility and benefiting both researchers and publishers.
Improved Customer Experience
Conversational discovery offers a user-friendly experience, akin to interacting with a person who understands questions and intentions, memorizes previous information, and provides answers in preferred formats. This method, called flipped interaction, suggests related or follow-up questions to refine users' thought processes, delivering answers with evidence in rich formats like links, images, audio, and videos.
Better Content Engagement and SEO
Enhanced discoverability leads to increased views and engagement, benefiting monetization through ad impressions, subscriptions, or other revenue streams. GenAI directly optimizes SEO and improves search experiences, encouraging content sharing and linking, which boosts search engine rankings.
Addressing Pain Points in Research and Publishing
GenAI can suggest relevant journals for research papers and reviewers, though experiments show it often recommends top-tier journals due to training data bias and struggles with accurate reviewer suggestions. Dedicated journal and reviewer suggestion services trained on specific data outperform generic GenAI applications, providing more detailed results.
The Future of Discovery Solutions and Methodologies
Discovery methodologies are evolving, from traditional keyword and Boolean searches to neural semantic search and Retrieval Augmented Generation (RAG), integrating updated knowledge with LLMs to generate comprehensive answers. AI agents, representing systems that perceive, decide, and act autonomously, will drive future search and result generation, leading to personalized, rich-format answers on demand.
As search solutions and methodologies advance, discovery aims to achieve business or task objectives. Transitioning from procedure-oriented to objective-oriented discovery focuses on defining clear goals and automating tasks, exemplified by Google's multistep reasoning capability.
In conclusion, GenAI-powered tools are revolutionizing information discovery, providing personalized, efficient, and accurate solutions for researchers and publishers, ultimately enhancing the scholarly research landscape.
