How AI Can Be an Ally for Content Design

---

Ever wondered how you could harness the magic of AI to supercharge your content design process? If so, you're not alone. Just last week, I had the opportunity to dive deep into this fascinating topic during a Pep Talk with The Content Club. The energetic participation and insightful feedback truly added to the richness of the discussion, and inspired me to weave together this blog post. This is not necessarily a more in-depth or comprehensive take, but a retelling of that tale, touching upon the same themes and points we explored in the presentation.

Generative AI has a bad rep in the content community, and unfairly so‌. It’s not the monster it’s been made out to be. For content design, and certainly for me, it’s been like discovering a bit of magic. It’s not here to steal our jobs, it’s here to supercharge us and give us more room to focus on what we truly love — creating meaningful, user-centred content.

Whether you’ve been exploring the benefits of AI for a while, or just beginning your journey, let’s discover how AI can lend a magical touch at each step.

TL;DR (too long; didn’t read summary)

AI is transforming the landscape of content design, acting as a magical ally rather than a threat. It can assist in every step of the content design process, from discovery to testing and iteration. Despite potential pitfalls such as data privacy concerns and 'hallucinations', responsible and informed use of AI, coupled with evolving regulations, can make it a powerful tool for content designers. Remember, while AI can assist us, the real magic lies in human creativity and empathy. Let's embrace AI, and let it enhance, not replace, our work.

---

Demystifying AI

Generative AI, like OpenAI’s ChatGPT, uses a type of AI technology called Natural Language Processing (or NLP, for short). NLP is all about understanding, analysing, and producing human-like language. It's trained on nearly everything from the web up until the end of 2021.

Using tools such as ChatGPT, Google Bard, Perplexity AI (and too many others to list) involves giving a 'prompt,' your starting point or instruction to the AI, much like starting a conversation.

Surprisingly, you might already be using AI without realising it, for instance, predictive text on your keyboard — that's AI! Soon, it'll become an integral part of our design tools as tech giants like Google and Microsoft have AI features coming very soon.

Crystal ball of discovery

So kicking off at the beginning of the content design process, AI is like a crystal ball, uncovering hidden insights. For instance, use AI to analyse the most common terms on your website to help you understand what your users are looking for. However, if you're using AI tools like ChatGPT for research and wish to cite your sources, ensure they exist to avoid 'hallucinations' (we'll touch upon this later), or consider Perplexity AI, specifically designed for this purpose.

Prompt examples:

• Help me identify key themes from these user interviews.

• Generate a summary of common user queries from these chat logs.

For other research, that’s where I’ve had lots of fun experimenting. Using the Scraper plugin on ChatGPT (available on the Premium plan), I fed it URLs from the World Economic Forum's The Future of Jobs Report 2023 and asked how it thinks the role of a content designer will evolve based on that report. 

Remember, though, humans are notoriously bad at predicting the future, so we can't entirely rely on AI's predictions either. That being said, it’s intriguing to imagine the exciting possibilities that the future might hold for us.

Wand of defining user needs

AI is like a magic wand, empowering content designers with enhanced creativity and productivity. It can generate ideas based on the research you’ve gathered in discovery, propose potential user personas and unlock new realms of content possibilities.

Prompt examples:

• Based on the user research, define the top 3 user needs that our content should address.

• Generate user personas based on the feedback from our customer help centre.

Fairy Godmother of ideation

AI is like your Fairy Godmother, magically transforming your content to meet diverse user needs. It helps you generate an array of ideas, regardless of the circumstances.

Prompt examples:

• Suggest multiple layout ideas for an online vehicle registration web page.

• Propose different ways to display heart rate data on our health mobile app.

• Brainstorm ideas to structure a new starter guide for our software’s help centre.

Magic mirror of design

AI is like the magic mirror, reflecting our deepest designs in content design. It can suggest engaging and empathetic headlines, relevant keywords and language choices — all with accessibility at the heart.

Prompt examples:

• Draft an error message for users who entered the wrong email address.

• Review this copy for tone consistency and suggest improvements.

• Analyse this page’s content for readability and engagement then suggest improvements.

Inclusivity is a major draw for content designers using AI, allowing you to personalise content to individual needs and preferences, improve accessibility by generating simple language, and promote inclusive design by training AI to understand and respect the nuances of different cultures and abilities.

Potion of testing and iteration 

With AI you can continuously refine your content. It can also act as a vigilant guardian of your brand, suggesting edits to maintain your brand’s voice and tone as well as checking your grammar and spelling.

Prompt examples: 

• Based on user feedback, suggest revisions for the instructions on how to apply for a driving licence.

• Proofread the in-app messages in our mobile app for language clarity and motivational tone using UK English.

• Check my content is consistent with the GOV.UK Design System and suggest what changes are needed in bold and explain why.

Wise counsellor

AI can be like a wise counsellor providing guidance and advice that’s tailored to your needs to support your development. From giving you feedback on your CV, and writing the perfect cover letter, to learning a new skill — it can help you grow personally and professionally.

Prompt examples:

• I want to be promoted. Here is a list of my responsibilities and new responsibilities I have just started. Help me write a proposal for a new role.

• Here’s a list of my job history, can you suggest potential career options for me to consider?

• I want to learn how to write a romance novel, provide me with a 12-week learning plan.

Potential pitfalls

Many are concerned about the ‘hallucinations’ you get with any AI tools. This is when the AI generates a confident but incorrect response when it lacks the information it was trained on. This hasn’t been an issue with me with the tasks I’ve been doing, but it’s a concern nonetheless. Many are also concerned with risks of intellectual property (IP) risks and concerns about data privacy, due to usage that gives companies like OpenAI (the owner of ChatGPT) permission to use information in future language model training. 

Data and privacy concerns

Here are a few things that might help reassure you:

• You can avoid all of this if you turn off your chat history.

• If you don’t want to do that, consider using an API. This is a set of rules that allows your application to communicate securely with the AI system. It can provide instructions to the AI about how to handle your data, like storing or remembering it, enhancing the privacy of your information.

• Most of our ideas are not unique. Most text-heavy work doesn’t contain true IP - but proprietary information or code, personal details, or customer information is a different story. 

• In the words of Rachel Woods, apply the ‘Reddit test’ - if you wouldn’t post something anonymously on Reddit, don’t put it into AI.

If you’re still not convinced, consider a tool like Writer. Unlike other language models, is based on your own company’s data, keeping your data private so it’s never used in model training.

Bias in AI models

AI models may learn and replicate biases present in their training data, which is a major concern for AI ethics. Addressing this involves:

• Diverse data collection: Gathering inclusive, diverse data can help reduce biases.

• Regular bias checks: AI systems should be regularly audited for biases during training and deployment.

• Transparency: Making AI decision-making processes more understandable can help spot biases.

• Bias mitigation: Techniques are applied to data before or after model training to reduce bias.

• Public input: Involving users or those affected by AI in influencing system rules and policies to promote fairness (Time, 2023).

Environmental impact

Adding generative AI to search processes could require four or five times more computing per search and the current infrastructure of data centres may not be sufficient to cope with this increase in demand​. The energy requirements for such operations could contribute to the world's greenhouse gas emissions (Wired, 2023)​.

However, its contribution is relatively small compared to sectors like transportation or textiles​​. AI systems are also likely to become more energy-efficient. Google is already releasing a version of Bard that consumes less energy, and tech giants like Microsoft and Google have committed to reducing their carbon emissions and transitioning to renewable energy sources (Wired, 2023).

Looking to the future

Working together we can work alongside AI, using it as a tool for good.  Researchers and organisations are working actively on responsible AI development, ethical considerations, and regulations and the EU passed the landmark AI regulation bill on June 14, 2023.

This is the world's first major set of comprehensive rules regulating AI technology. The bill includes provisions for the use of AI in areas such as facial recognition, social scoring, and algorithmic decision-making.

Google, Amazon, and Microsoft back the EU AI Act. Saying that it would help to ensure that AI is used responsibly and ethically. The companies have also said that they are willing to work with the EU to implement the act (Washington Post, 2023). So let’s keep advocating for ethical AI and engage where we can to influence policies that serve and protect us from the pitfalls of AI.

7 tips to craft the perfect prompt

Just like having a team of elves, AI works tirelessly in the background to provide help when you need it. However, AI cannot act or think independently and specific instructions are needed to get an answer that meets your expectations. Here are my top tips.

1. Start your prompt with “Act as an expert in…” that will give the AI a frame of reference to start with.

2. Define the style and tone: if it’s simple language you’re after, or you want to refer to a style guide, make that clear.

3. Provide the context: what’s the problem you’re trying to solve, who’s it for?

4. Specify the content type: is it for a website, a presentation, a blog post etc.

5. Iterate as you go: Don’t expect perfect results from the first prompt, just like the iterative process in design, the same is true for interacting with AI.

6. Fact-check: It’s essential to fact-check as you go, it’s a language model so critically evaluating the content of the output is vital. 

7. Learn how to write better prompts: Prompt engineering is the latest buzzword in all things AI. This involves learning how to write effective prompts. And that’s what we’re all learning — It’s not a job, it’s a skill and practice makes perfect. The ONE ChatGPT Prompt to Rule Them All is one of the best ways that’s helped me practise.

Continue your journey of AI discovery

To learn more, here are some of the latest free learning resources:

• Google’s Generative AI learning path

• Microsoft and LinkedIn’s AI Skills Initiative

• Udemy ChatGPT courses

Final thoughts

Remember, AI tools can help you in your writing process but they aren’t perfect and often won’t produce what you’re looking for on the first try. But be patient, iterate and refine your approach to get the most out of these tools. 

While AI can be a powerful ally, the true magic of content design lies in human creativity, empathy and understanding. In the words of Dumbledore (and humour me for a second in the absence of content design quotes in the theme of magic and fairy tales.)

“Words are, in my not-so-humble opinion, our most inexhaustible source of magic.”

— Dumbledore

So let’s embrace AI’s assistance while recognising the indispensable role of human expertise. And let’s take the opportunity to take the lead role as experts in language and UX to demonstrate our value to support the business in leading AI implementation. This is a massive opportunity for us, and who better to do it than content designers?

---

References:

1. YouTube, 2023. Pep talk Kate O'Brien: How AI can be an ally for content design. [video] Available at: https://www.youtube.com/watch?v=BeDZHKn0Yng [Accessed 02 July 2023].

2. OpenAI, 2023. OpenAI Chat. [online] Available at: https://chat.openai.com/ [Accessed 26 June 2023].

3. Google, 2023. Google Bard. [online] Available at: https://bard.google.com/ [Accessed 26 June 2023].

4. Perplexity AI, 2023. Perplexity AI. [online] Available at: https://www.perplexity.ai/ [Accessed 26 June 2023].

5. World Economic Forum, 2023. The Future of Jobs Report 2023. [online] Available at: https://www.weforum.org/reports/the-future-of-jobs-report-2023/ [Accessed 27 June 2023].

6. UK Government, 2023. Design System. [online] Available at: https://design-system.service.gov.uk/ [Accessed 27 June 2023].

7. OpenAI, 2023. OpenAI. [online] Available at: https://openai.com/ [Accessed 27 June 2023].

8. Woods, R., 2023. LinkedIn Profile. [online] Available at: https://www.linkedin.com/in/woodsrach/ [Accessed 27 June 2023].

9. Writer, 2023. Writer Platform. [online] Available at: https://writer.com/platform/ [Accessed 27 June 2023].

10. Time, 2023. OpenAI EU Lobbying AI Act. [online] Available at: https://time.com/6288245/openai-eu-lobbying-ai-act/ [Accessed 27 June 2023].

11. Wired, 2023. The Generative AI Search Race Has a Dirty Secret. [online] Available at: https://www.wired.com/story/the-generative-ai-search-race-has-a-dirty-secret/ [Accessed 27 June 2023].

12. The Guardian, 2023. Techscape European Union AI. [online] Available at: https://www.theguardian.com/technology/2023/jun/27/techscape-european-union-ai [Accessed 27 June 2023].

13. The Washington Post, 2023. EU Parliament Approves AI Act. [online] Available at: https://www.washingtonpost.com/technology/2023/06/14/eu-parliament-approves-ai-act/ [Accessed 27 June 2023].

14. YouTube, 2023. The ONE ChatGPT Prompt to Rule Them All 👑 [video] Available at: https://www.youtube.com/watch?v=OgYQAS9LY3o [Accessed 21 June 2023].

15. Google, 2023. Cloud Skills Boost. [online] Available at: https://www.cloudskillsboost.google/paths/118 [Accessed 02 July 2023].

16. Microsoft, 2023. AI Skills Resources. [online] Available at: https://www.microsoft.com/en-us/corporate-responsibility/ai-skills-resources [Accessed 02 July 2023].

17. Mashable, 2023. ChatGPT Free Courses. [online] Available at: https://mashable.com/uk/deals/chatgpt-free-courses [Accessed 02 July 2023].

18. Rowling, J.K., 1997. Harry Potter and the Deathly Hallows. 1st ed. London: Bloomsbury, p.723.