How to Harness AI in Publishing – an Overview
The PTC’s webinar on 25 June discussed how publishers can navigate the latest developments in generative artificial intelligence (AI). Three expert panellists explained what’s happening in marketing, new-product development and the legal side of AI.
We’ve highlighted key insights below.
Winning the book-marketing game with AI
Book marketing can be tough, explained Brooke Dobson, Co-Founder and Chief Commercial Officer of Shimmr AI, with media fragmentation, diminishing attention and marketing waste that low-margin books can’t sustain. Adopting an AI-first mindset can help overcome these challenges – by using it for research and strategy, creative ideation, asset production, testing and optimisation, metadata and search-engine optimisation.
For most publishing workflows, Claude (Anthropic), Gemini (Google) and ChatGPT (Open AI) are the most reliable.
💡Top tip: Use AI to automate the tasks you dread and to enhance those you like. Experiment with as many tools as possible.
How to work with an AI
-
- Start simple, using a conversational tone
- Give context and describe your needs clearly
- Be specific – detail your audience, purpose and desired outcome
- Guide with chain-of-thought instructions to clarify complex tasks
- Request 50 ideas, not 5, and push it to elaborate on promising concepts
Brooke suggests these time-saving tools:
▶️ Otter for transcription
▶️ Intercom for customer support
▶️ Gamma to create presentations
▶️ Veed to make videos.
To inject more creativity into your marketing, try:
▶️ Flux to create storyboards and ads
▶️ Hailuo for animations
▶️ ChatGPT/DeepSeek to create drafts, blurbs, metadata and emails.
💡Top tip: Choose the right model for your purpose – whether fast, powerful or specialist – and disable the training option for privacy.
Win on Amazon with AI
Amazon’s A10 algorithm prioritises external traffic and rewards relevance, trust and sustained reader satisfaction. AI gives you the tools to deliver these – faster, smarter and at scale.
Sustainability
Brooke advised contributing to sustainability by using low-energy models, equivalence metrics – such as comparison with car mileage – or more balanced technologies.
How to create trusted, safe pedagogical AI solutions
Eirik Wahlstrøm, Co-founder and CEO of Ludenso, began with a quick poll, showing that many of us use AI once a week or more. He highlighted that AI now creates more internet content than humans, and that it’s tricky to distinguish between the two. How can we trust what we see and hear? Navigating the AI landscape isn’t easy and publishers have big challenges to tackle, including:
-
- ensuring accuracy and quality
- safeguarding Intellectual Property (IP) rights
- build or buy?
- choosing the right AI models
- reskilling and upskilling
- meeting elevated customer expectations.
How to build accuracy and trust
Publishers can be the trusted option for pedagogical AI use with Retrieval Augmented Generation (RAG), a technique that optimises the output of large language models (LLMs) by integrating them with external knowledge sources. Ludenso’s LAIA is one of several RAG models – including NoteBook LM (Google), Custom GPTs (Open AI) and Haystack (Deepset). The LAIA AI assistant shows how a RAG model can deliver validated, trusted, accurate content to match your audience’s needs and deliver a personalised, branded experience. For example, it can offer:
-
- students tailored exams, homework or in-class questions
- editors generation of multiple-choice questions, lesson plans or PowerPoint presentations.
The AI assistant can be tuned for age level, subject area, grade level, curricula goals, tone of voice and language. You can add features such as quizzes, flashcards and concept explanations, plus ‘Socratic’ modes versus giving away the answers. Data is secured locally with LLM plug-ins.
To get started, Eirik recommends these simple steps:
-
- Test several different models and variants, and ensure they are not being used as a training model.
- Make sure you’re not giving away IP in the process.
- Share your findings with peers and build confidence.
💡Top tip: Ensure AI applications are validated and include humans in the loop.
How to protect copyright in the age of AI
Copyright and AI is complex, as Leslie Lansman, Head of Permissions & Licensing at Springer Nature and Chair of the Public Policy Committee, the Association of Learned and Professional Society Publishers (ALPSP), explained.
Leslie’s list of (non-exhaustive) considerations will help you choose the right AI tool.
She advises that you carefully examine any terms, conditions or other requirements in accessing or using an AI system. Understand how the tool works and what happens to your material if you stop using it.
-
- Create policies and training on how to use your AI tools. Establish rules and/or technical solutions to ensure data privacy, confidentiality and copyright compliance.
- Understand the security and resilience of your AI tools.
- Use content in the public domain or under appropriate licence. Questions remain on how best to deal with the attribution requirement if using Creative Commons BY material.
- Avoid using protected works, copy/pasting or describing styles, images or logos. See Copyright Licensing Agency AI Licences for AI workplace licences that allow limited reuse of participating publishers’ works in some internal AI systems.
- Identify the purpose and scope of your AI usage. Add your unique touch. Ensure a human is thoroughly reviewing and verifying any output.
- Address to what extent you will disclose AI use. Explain what AI you used and how.
- Document your AI usage and monitor for compliance. Evaluate the AI tools.
- Think about sustainability!
Brooke Dobson
Brooke Dobson is Co-Founder and Chief Commercial Officer at Shimmr AI (Philadelphia, USA). Follow her on LinkedIn.
Eirik Wahlstrøm
Leslie Lansman
Leslie Lansman is Head of Permissions & Licensing at Springer Nature, Chair Public Policy Committee at ALPSP (London, UK). Follow her on LinkedIn.