From maintaining research integrity to diversifying revenue sources, publishers are facing an increasingly broad set of challenges, needs, and goals, especially in the open science movement. As artificial intelligence (AI) begins to play an ever-bigger role in the scholarly publishing landscape, how might it help solve some of these pain points?

In this post, I’m diving into ten of the biggest challenges faced by publishers — based on my experience — and exploring how AI solutions have the potential to help.

1. Enhancing content discovery and dissemination

Suffering from information overload? You’re not alone… With so much information out there, users need to accurately – and quickly – get answers and the information they need. At the same time, publishers need to effectively disseminate their publications to improve visibility and make content easier to find.

To increase traffic and keep readers engaged for longer, publishers need to boost content visibility and accessibility while creating an immersive user experience.

So, where could AI help with research publishing? First, AI can enhance discoverability beyond keyword searches with contextual understanding, semantic search, and natural language Q&A. It can also help deliver more personalized content recommendations (based not only on interests, but on funding, institutional policies, etc.), while making images, video, and other interactive digital content as easily discoverable as an article. This enables wider, deeper, and faster multimedia information discovery in the global research landscape.

AI technologies will also enable intelligent content enrichment, metadata extraction, and voice search, making it much easier for researchers to get the information they need quickly. In fact, as we move from traditional keyword discovery to conversational discovery, content discovery is one of the areas most impacted by generative AI (GenAI).

2. Modernizing publisher operations

Many existing journal workflows (submission, peer review, production, etc.) can be complicated, expensive, inefficient, and involve a lot of manual work. Publishers want to support editors and authors in making publishing more seamless while streamlining operations and ensuring that processes are fit for the needs of the open science movement.

AI could help with many of the common pain points publishers experience across different workflows – for example, by finding independent, relevant peer reviewers and editors. We could see AI-powered author services that include tools like author summarization or translation, automatic poster generation, and GenAI-powered copyediting services during production. AI can be applied during the submission phase (think auto submission and quality screening) and can be used in peer review to facilitate reviewers’ and editors’ work and speed up the decision-making process.

3. Improving research integrity

Concern for and interest in research integrity has increased significantly during recent years. AI can be used to reduce fraudulent publication by detecting many common integrity issues such AI-generated content, data falsification, problematic references, and researcher/editor/reviewer verifications. Many publishers such as Wiley, Elsevier, and Springer Nature have already invested heavily in this technology and have their own AI-powered integrity services.

GenAI has the power to boost research quality and collaboration, but it also brings challenges to research integrity, such as a potential increase in plagiarism, image manipulation, and paper mill content. Its ability to create human-like text and alter images can make fraud harder to spot. As research misconduct evolves, so too will the strategies to detect and stop it.

Importantly, while these AI tools are here to make people’s jobs easier, there is still very much a need for human oversight to make informed decisions. And transparency in – and the ability to explain – these AI services is key to building trust in the research community.

4. Diversifying revenue sources

Knowledge mining and discovery tools that leverage AI can generate revenue streams by selling more value-added knowledge, rather than just content. They also offer domain-specific paid knowledge services to create connections between people and people, and between people and knowledge. In the era of GenAI, publishers can diversify revenue by generating and selling new versions or formats of content – for example, existing content in new languages and rich multimedia content. Plus, human-curated content is and will continue to be highly valued.

5. Author-centric publishing

Authors frequently encounter bad experiences on their research journey, including inefficient authoring, submission, and transfer workflows. At the same time, there is a lack of new services to better find and retain authors, and to serve them better.

For publishers, we know that attracting more authors and high-quality content are key in the competitive open access environment. Improving user experience and author satisfaction are critical to success. AI can offer solutions including:

• Intelligent authoring – including auto summarization, term/format checks and reformatting, and language improvement
• Smart author services – including auto poster generation and abstract creation
• Auto submission and transferring
• A more personalized research journey with increasingly sophisticated user profiles
• Expert/collaborator suggestion tools to link global experts and talents
• Auto translation to improve accessibility and readability for global communities

6. Content management and monetization

As we move into an open future, research output has expanded to include not just the article, but code, data, video, and more. At the same time, it has become increasingly costly and complex to continuously publish every version of an article plus its associated outputs.

Legacy content is often of great value to researchers; however, users can’t easily access it since most historical content is made up of printed papers or scanned images. Publishers need to efficiently manage content and translate research for impact and application beyond the scholarly journal model. This includes managing and supporting discovery for all research outputs beyond the article.

AI has the potential to play a huge role here, through personalized content collection and advanced multimedia content enrichment, such as cross-language translation and audio content creation. AI can also be used to manage copyright by automatically tracking and checking rights and detecting copyright infringement and misuse. Moreover, as STEM becomes increasingly multidisciplinary in nature, more effective cross-disciplinary content classification and management are needed.

7. Audience management and monetization

Finding the right audience – and understanding them — is challenging. Serving up ads to the wrong people can result in negative consequences for brands, as well as low engagement and click-through rates. And while marketers may hope to capture a snapshot of a user’s online engagement to monetize it via ads, they may not ‘follow’ users across their online engagement journey. Overall, there is a lack of information that would identify users across their journeys and turn them into monetizable audience segments, where permissible under applicable law.

There needs to be broader communication, connections, and transparency among publishers, researchers, society, and consumers, attracting new audiences and providing actionable and engaging experiences.

AI can help to create a unified, consistent way to identify and segment audiences that leverages publishers’ superior knowledge of users and their interests as derived from engagement on multiple channels, allowing integration in existing workflows to improve audience targeting solutions, in compliance with legal requirements. For example, Hum is a customer data platform (CDP) which builds for content-rich organization. Atypon has also recently announced the development of the CONNECT CDP. These platforms all enable scholarly publishers and societies to enhance their audience understanding and targeting capabilities.

8. Insight and research analytics

Confusion and disagreement in academia over what should be measured — and how it should be measured — creates ambiguity in the publishing marketplace. Publishers want to expand, enrich, and link their content with worldwide research information, and compare local usage patterns to global patterns across publishers. AI can help to assess and predict impact, trends, and coverage to help better identify gaps and opportunities. GenAI also empowers publishers to leverage intelligent business analytics, enabling near real-time decision-making by simply asking natural business questions, without the need for data scientists to craft complex SQL queries.

9. Cybersecurity and identity management

Cybercrime is a huge threat to the entire scholarly ecosystem, and safeguarding data and privacy is crucial. While publishers want to better serve users by understanding their intentions and interests, it is hard to achieve this in the current environment in which anonymous users have grab-and-go behaviors. Users must repeatedly enter their information to register different accounts, and it’s hard to know whether they’re eligible to access some content or APC waivers based on the affiliation provided. It is better for publishers to provide seamless end-user access to content with single sign-on (SSO) privileges.

AI has the potential to automatically identify and disambiguate researchers’ and institutions’ information based on attributes such as names, publication history, ORCiD, ROR/Ringold ID, etc. to ensure more accurate and intelligent identity management, where permitted and conducted in compliance with applicable legal requirements. AI can also help with intelligent bot detection, which identifies and flags any suspicious activity in usage event data.

10. GenAI for publishers and societies

Mission-driven organizations like scholarly societies have concerns about the ethical deployment of AI and bias. And while some publishers think that it’s simple to just hire people to build or buy in services from other vendors, others think AI is too hard to develop and govern, especially when it comes to copyright infringement and handling data ethically.  As with any nascent technology, many questions remain: no reliable estimates exist on the extent of unauthorized use of content to train AI models; publishers have yet to assess the impact of GenAI on the shape of their workforce; and some are still working to establish their AI principles and policies.

In conclusion, AI is increasingly aligning with the open science principles of transparency, accessibility, and collaboration, helping publishers navigate the challenges and pain points discussed above. As we continue to advance as an industry, with large language models (LLMs) becoming more powerful and intelligent, we can anticipate that even more innovative AI-powered solutions will emerge, with startups and vendors joining the landscape. The rise of standalone, cloud-based AI services offers flexibility, allowing seamless integration into existing workflows and enhancing the publishing process.

This evolving ecosystem and the lowering entry barrier of AI technology underscores the importance of platforms that integrate AI solutions and deliver value effectively to customers, creating greater efficiency. However, as AI capabilities rapidly evolve, governance is struggling to keep pace, making trust, risk, intellectual property compliance, and security management critical. We can also expect the regulatory landscape to expand, as new guidelines and frameworks are established to ensure responsible AI development and application. The future holds immense potential for AI in scholarly publishing, but with it comes the need for thoughtful oversight to maximize its benefits.