Fairpatterns’ Post

🔍 AI Roadmap to Avoid Dark Patterns in Digital Marketing 🔍 In today's digital age, it’s crucial to steer clear of dark patterns in AI-powered marketing. Fairpatterns has developed a comprehensive AI Roadmap to help digital pioneers navigate these challenges ethically. Here’s why and how to avoid these deceptive practices: Why? 🛡️ Avoid Brand Damage: - Unethical AI, riddled with dark patterns, can backfire on user trust, damaging your brand reputation and value. - Users are becoming more aware of dark patterns. Being caught can erode trust, impacting customer loyalty and brand reputation. - Dark patterns create a negative user experience, leading to cart abandonment, unsubscribes, and lost revenue. 📖 Tell a Story You’re Proud Of: - Certain groups are especially vulnerable, including minors, the elderly, and low-income users. Ensuring your AI respects these users fosters trust and inclusivity. 🌟 Create Opportunities, Not Harms: - Transparency and fairness generate more long-term revenue than manipulation and deception. Ethical practices avoid hefty fines and build lasting customer relationships. How? 1️⃣ Step 1: User Understanding & Empathy - Conduct in-depth user research to understand their needs and vulnerabilities. - Educate your team on cognitive biases and how dark patterns exploit them. - Develop an Ethical AI Manifesto focusing on user autonomy, fairness, and transparency. 2️⃣ Step 2: Crafting Responsible AI Experiences - Use tools that prioritize user consent and clear explanations of AI recommendations. - Empower users with control over their data and interactions. - Design AI-powered experiences with accessibility in mind. 3️⃣ Step 3: Prompt Engineering for Ethical AI - Frame your prompts around user benefits, not manipulation. - Avoid manipulative language and be transparent about AI's role in your campaigns. 4️⃣ Step 4: Continuous Monitoring & Improvement - Use A/B testing to promote informed choices and avoid manipulative tactics. - Actively solicit user feedback and conduct regular audits to keep your AI ethical. - Incorporate Explainable AI (XAI) techniques and monitor for prompt bias. Embracing ethical AI-powered digital marketing not only builds trust but also ensures sustainable value for your brand. Let’s commit to transparency, fairness, and user-centric approaches to create a better digital world. 🌐✨ #AI #EthicalAI #DigitalMarketing #ConsumerTrust #DarkPatterns #UserExperience #Transparency #Fairpatterns

I was really interested to see that EMARKETER in their 'Hyper-Personalization Explainer 2024' paper went so far as to call AI 'essential' to hyper personalisation in retail, and in part that seems to be specifically around GenAI to scale up content creation. Bastian Hagmaier you raise the point about GenAI from a privacy and legal perspective, but there's also a question about whether it's the right thing to do at all - brand distinctiveness is one of the most powerful levers a brand can pull to impact long term performance. GenAI content is by nature regression to the mean. Are we valueing convenience over efficacy? - can we just whip up mediocre marketing faster than ever before? The focus of AI in marketing should be to free up marketers time and energy to be even more creative, to create value for their brands through distinctiveness that they would not achieve spending hours each week trawling data for opportunities, building reports to justify action and then building segments etc - these are the things that AI does do effectively. When I first joined Emarsys we called this 'creative renaissance': using AI to free the marketer to be a creator. (what today we'd call Power to the Marketer) What do you think about that?

Recently, I've been asked a lot about #AI by our customers, as well as how I see the mid-term future of AI products for Marketing. I'd like to share some of the most important points I often discuss, focusing on three main areas influencing how marketers use AI today: 1. Key Types of AI for Marketers – 3 examples Language Models (LLMs): LLMs are evolving incredibly fast. They can provide fantastic assistance in condensing thoughts, changing tones, and brainstorming, just to name a few examples. However, they are often unreliable with facts. For brands, it's crucial to manually ensure the generated content aligns with the expected tone and message. Pattern Recognition Algorithms: Pre-trained models can uncover insights in your data that are nearly impossible to detect manually. These algorithms help identify new customer segments, making targeting more successful. They also aid in decision-making by highlighting actions that lead to very successful or unsuccessful outcomes. Image Generators: These tools allow you to quickly and easily create, edit, or augment images for your marketing campaigns. Precision in your descriptions is key to getting the desired results, making the editing process more flexible and efficient. 2. Regulations and Uncertainties From copyright laws to privacy concerns, significant uncertainties around AI hinder many marketers from fully utilizing these technologies. Key questions arise: - Who owns the text generated by an LLM? - Who owns the images it creates? - What data was used to train these models? - Under GDPR regulations, what data can you share with a pattern recognition algorithm without additional consent? - etc etc...! Moreover, if AI interacts directly with customers (e.g., through personalized emails or push notifications), how do you ensure it doesn't damage your brand's reputation by hallucinating or downright offending your customers? 3. Feature, Not a Product The landscape of AI regulations, the tools that are already out there, the privacy controls around it and the reliability of these systems are continuously changing, and in most cases improving drastically every day. Reality is, you could go and use so many of these tools *today*. But most Markters *won't*. Why? Because it's either too difficult, too uncertain, too hard to stay on top, too many changes of what can and can't be done. Sure, a lot of Marketers have already successfully started to use AI, but there's a long way to an integrated main-stream adoption. That's why what we believe in making AI a reliable and trusted part of our Platform in a way that allows you to actually use it instead of having to think about it. That way, it becomes an essential feature that integrates directly in your workflow and extends your capabilities without needing to worry about all of the implications. #PowerToTheMarketer Discover what the SAP Emarsys platform offers today — and we're just getting started! https://bit.ly/3KUCGkc

Have you heard about Meta AI’s latest breakthrough? Allow me to introduce Llama 3.1, the AI that's set to redefine digital marketing. Here's why marketers should be excited: 1. Lightning-Fast Data Processing 🚀 In marketing, time is of the essence. Llama 3.1 isn’t just faster—it’s a powerhouse! Imagine cutting your campaign analysis time in half, or even more. Whether you’re managing large datasets or running complex algorithms, Llama 3.1 handles it with unmatched speed and precision. This means quicker insights, faster optimizations, and a significant boost in campaign performance. 2. Superior Customer Understanding 🗣️ Ever wished your AI could understand and converse just like a human? Llama 3.1’s natural language processing is incredibly nuanced. It doesn’t just follow commands—it gets the context. This means more intuitive chatbots, personalized content, and interactions that leave your audience engaged and satisfied. 3. Effortless Scalability 📈 As your marketing campaigns scale, Llama 3.1 grows with you. Its seamless integration capabilities mean you can expand your AI usage without missing a beat. Whether you’re entering new markets or ramping up your content production, Llama 3.1 scales effortlessly alongside your ambitions. 4. Ethical and Responsible AI 🤝 In a world where AI ethics are paramount, Meta AI has set the bar high. Llama 3.1 is designed with robust ethical safeguards, ensuring fair and responsible use. This commitment not only builds trust but also aligns with our goal of creating marketing strategies that resonate authentically with audiences. Real-Life Magic with Llama 3.1: 🌟 Hyper-Personalized Campaigns: Picture an AI that understands your customers as well as you do. Llama 3.1’s empathetic responses and deep understanding allow for hyper-personalized marketing campaigns that truly resonate with your target audience. 🌟 Content Creation on Steroids: Need compelling blog posts, social media updates, or ad copy? Llama 3.1 can generate high-quality, engaging content in no time, letting your team focus on strategic, creative tasks. 🌟 Insightful Data Analysis: Dive deep into your marketing data and uncover actionable insights faster than ever before. Llama 3.1 turns raw data into valuable intelligence, driving smarter marketing decisions and more effective campaigns. 🌟 Breaking Language Barriers: Expanding globally? Llama 3.1’s multi-language support ensures you can communicate effectively with audiences around the world, making international campaigns smoother and more effective. The future of marketing is here, and it’s powered by AI like Llama 3.1. This isn’t just about technology—it’s about transforming how we connect with and understand our audiences. I’m excited to see how Llama 3.1 will revolutionise our marketing strategies and drive innovation. Have you explored Llama 3.1 yet? I’d love to hear your thoughts and experiences! Drop a comment below #MetaAI #Llama3.1 #FutureOfMarketing #AIMarketingByNSK

So I was doing some research for a 2025 Predictions piece I wrote for the Capitol Communicator and was doing a bunch of thinking about ad tech and generative AI and how the two might come together in new ways. So a few ideas popped into my head: 1. What about using generative AI to auto-generate hyper-targeted content on the fly based on user profiles? Rather than using AI to generate large chunks of content, what if a CMS with generative AI capabilities could generate a custom article for each reader by automatically editing an existing piece? 2. In a similar way, what if AI could be used to power customized social media content targeted to individuals rather than audience segments? I know this has all sorts of legal and ethical problems, but it seems like as long as they identified themselves as such, AI could be used to create personalized "Brand Buddies" based on specific brands that could act as personalized "friends" to those who wanted to opt-in to them, learning our individual preferences and uses and suggesting new ways to use a brand or providing individualized support. For example, if an REI or GoreTex "brand buddy" might learn from my social activity where I liked to go hiking or where I'm planning on taking a trip and could assist me with making sure I had the right gear or provide specific suggestions for how to get the most out of what I already owned. 3. The language processing capabilities of AI could also be deployed as a "share shield," a bot that would check everything shared with you on social media or even in your texts, which would provide customizable fact-checking on news stories, memes, or other online flotsam people like to share over these channels. It could potentially even auto-respond to people "sharing" non-fact-based materials with links to reputable sites that refute their bogus claims and, based on preferences, even block people if they don't stop sharing BS content. Something like this could even potentially help shield us from the flood of mis/dis-information coming at us from other bots. Of course, flooding the net with more AI-generated content might not exactly be a good idea, especially if it's not labeled as such. Maybe creating a system where sites that ONLY use humans to generate content would be a good idea, too? Would you be interested in seeing a "Verified Human-created Content" certification?

The first time I used an AI detection tool, it painted almost all my content in angry red and returned a snarky comment: "Reads like AI." I fought against it until I couldn't. So, I decided to lean into it. I started writing in a way that the AI tool would accept as "human" (whatever that means, lol). Today, this finicky tool passes over 90% of my content as human-written (though not ideal, certainly an improvement). If you're in the same boat, here's what you can do: 💡 Use "You" or "Your": This simple trick makes sentences sound less robotic and more conversational. Instead of saying, “Utilizing advanced analytics tools can be a valuable way to improve a content marketing strategy,” try, “You can use advanced analytics tools to improve your content marketing strategy.” This is not always possible, but I try to sneak this in wherever I can, as long as the content stays concise and fluff-free. 💡Avoid Words Like "Streamline" and "Leverage": These are ChatGPT's go-to buzzwords and WILL get flagged by AI detection tools. Plus, they make your writing sound stuffy and dry. I avoid them whenever possible, opting for clearer and simpler language instead. 💡Use Bullet Points: This is a favorite hack of mine. When explaining something technical or complex, I break it down into bullet points. For some reason, AI detection tools consider bulleted content to be human-written. Plus, bullets improve clarity, so it’s a win-win. 💡Add Conversational Segues: Instead of the mundane "for example," try phrases like "a great example of this is..." or "Think about how this can..." Not only do they convince AI detection tools that your writing isn’t robot-generated, but they help bridge the gap between paragraphs, making your content flow more smoothly. Passing content through an AI detection tool feels like going through the wringer. I don’t know about you, but I always hold my breath while waiting for it to judge the content I’ve poured my effort into. I might not fully understand how these tools work, but I do know that I don’t want to spend extra time rearranging what I’ve already written just to "humanize" it. #contentwriting #AI #AIdetectiontools

We tried nearly all of the well-known AI detectors on the market right now, and this is what we found... … … … They’re all unreliable. (Womp womp) AI detectors, while they can be spectacularly accurate, they can be spectacularly wrong just as often, and you'll never know which result you're getting. Heather tested several industry leading detector tools, and each time, she ran several articles that she knew for a fact were 100% human written (once she even free-wrote the article directly into the tool) and a couple that she had ChatGPT create and copy/pasted into the checker verbatim. Maybe 50% of the time, it is accurate. Not necessarily that the content was written by AI, but that it contains basic, AI-style prose that should be adjusted. But when it's wrong, it's wrong. For example, Heather created 3 articles in ChatGPT and pasted them verbatim into one of the most widely used AI Detectors. The read on these three 100%-AI written articles were: 98% AI detected 61% AI detected 2% AI detected Then she ran an article that she knew was at least partially written by AI, and the checker returned a 0% AI score. No tool can actually detect AI. There is no invisible code between the words that AI writes, that these checkers can somehow see. AI checkers look for what it considers to be hallmarks of AI-written content, like sentence length and complexity, word choice and "randomness", human error like typos, and other proxy characteristics. The problem is that many of these proxies for AI-written content are also very similar characteristics of human-written SEO-focused content. Why? Humans write a certain way to appeal to the algorithm. We consider these to be SEO best practices for writing keyword-optimized content. ⬇️ This writing style works to get content ranked, so a large proportion of online content is written this way. ⬇️ LLMs scrape online content to "learn" information, including how to write. ⬇️ LLMs produce content that is heavily influenced in style, cadence, and word choice by SEO best practices. ⬇️ AI detectors use these trends in LLM-produced content as markers for detecting whether articles were written for AI. ⬇️ The LLMs' inputs are constantly changing an expanding, presumably allowing them to write in styles other than the boilerplate one that detectors are looking for, and this must be true because at least one of the well-known detectors has a "humanize" feature that allows its AI to produce more human-sounding content 😵💫. But SEO best practices stay largely the same — at least the ones AI detectors are looking at. ⬇️ AI Detectors mark AI-generated content as maybe all or maybe no AI, or maybe something in between, while writing content that was written to rank as maybe some or maybe all AI, even when a human write every word. Maybe these are helpful for professors grading papers, but for marketing content, and especially search-optimized content, save your budget and listen to your gut.

The Office of the Australian Information Commissioner has released two guides on AI. I have provided a summary of the Top 5 Takeaways from each guide below, alongside links to each guide: 🔷 𝗚𝘂𝗶𝗱𝗮𝗻𝗰𝗲 𝗼𝗻 𝗽𝗿𝗶𝘃𝗮𝗰𝘆 𝗮𝗻𝗱 𝘁𝗵𝗲 𝘂𝘀𝗲 𝗼𝗳 𝗰𝗼𝗺𝗺𝗲𝗿𝗰𝗶𝗮𝗹𝗹𝘆 𝗮𝘃𝗮𝗶𝗹𝗮𝗯𝗹𝗲 𝗔𝗜 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝑻𝒐𝒑 𝒇𝒊𝒗𝒆 𝒕𝒂𝒌𝒆𝒂𝒘𝒂𝒚𝒔: 1️⃣ Privacy obligations will apply to any personal information input into an AI system, as well as the output data generated by AI (where it contains personal information). 2️⃣ Businesses should update their privacy policies and notifications with clear and transparent information about their use of AI. 3️⃣ If AI systems are used to generate or infer personal information, including images, this is a collection of personal information and must comply with APP 3. 4️⃣ If personal information is being input into an AI system, APP 6 requires entities to only use or disclose the information for the primary purpose for which it was collected. 5️⃣ As a matter of best practice, the OAIC recommends that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. Link to the AI privacy guide: https://lnkd.in/gjsczBhi 🔷 𝗚𝘂𝗶𝗱𝗮𝗻𝗰𝗲 𝗼𝗻 𝗽𝗿𝗶𝘃𝗮𝗰𝘆 𝗮𝗻𝗱 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗶𝗻𝗴 𝗮𝗻𝗱 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗺𝗼𝗱𝗲𝗹𝘀 𝑻𝒐𝒑 𝒇𝒊𝒗𝒆 𝒕𝒂𝒌𝒆𝒂𝒘𝒂𝒚𝒔: 1️⃣ Developers must take reasonable steps to ensure accuracy in generative AI models. 2️⃣ Just because data is publicly available or otherwise accessible does not mean it can legally be used to train or fine-tune generative AI models or systems. 3️⃣ Developers must take particular care with sensitive information, which generally requires consent to be collected.  4️⃣ Where developers are seeking to use personal information that they already hold for the purpose of training an AI model, and this was not a primary purpose of collection, they need to carefully consider their privacy obligations. 5️⃣ Where a developer cannot clearly establish that a secondary use for an AI-related purpose was within reasonable expectations and related to a primary purpose, to avoid regulatory risk they should seek consent for that use and/or offer individuals a meaningful and informed ability to opt-out of such a use. Link to the developing and training generative AI guide: https://lnkd.in/g8CcsDbh . . . . #AI #ArtificialIntelligence #Privacy #DataProtection #OAIC #PrivacyGuidance #GenerativeAI #Compliance #DataPrivacy #EthicalAI #PrivacyPolicy #AIRegulation #DigitalEthics #Innovation #BusinessCompliance #TechForGood #AIResponsibility #AustralianLaw #PrivacyObligations

From now on, I'll be labelling my content as #AI, or not! I love AI and I love using new AI tools that allow me to do things that I couldn't possibly do without them, or to simply save me time. I use AI frequently for content development, but I have to say this is principally for planning content or finding alternative words, headlines or other components for what I write. I have nothing against using AI to create content - and, for example, I create a lot of images via AI - so long as the finished copy is fit for purpose. The new wave of #GenAI content tools can prove to be extremely valuable, but there are risks with mass producing content without proper scrutiny, in particular in the business world. Meanwhile - in my book at least - claiming that you wrote something because you requested an AI output, is not much different to briefing someone else to write your copy. The original idea may be yours, but you still didn't write the copy! I strongly believe that #transparency is the key to the ethical and responsible use of AI, and so I already try to label any AI-generated content that I produce as such. However, I'm now going to take this one step further and begin to use more explicitly labelling of my content. Here's my list of labels! [ 100% 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥 ] [ 100% 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥 + 𝘈𝘐 𝘐𝘮𝘢𝘨𝘦 𝘝𝘪𝘢 𝘔𝘪𝘥𝘫𝘰𝘶𝘳𝘯𝘦𝘺 ] [ 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥, 𝘌𝘥𝘪𝘵𝘦𝘥 𝘣𝘺 𝘈𝘐 ] [ 𝘈𝘐 𝘊𝘳𝘦𝘢𝘵𝘦𝘥, 𝘌𝘥𝘪𝘵𝘦𝘥 𝘣𝘺 𝘏𝘶𝘮𝘢𝘯 ] [ 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥, 𝘙𝘦𝘱𝘶𝘳𝘱𝘰𝘴𝘦𝘥 𝘣𝘺 𝘈𝘐 ] [ 𝘈𝘐 𝘐𝘮𝘢𝘨𝘦 𝘷𝘪𝘢 𝘔𝘶𝘴𝘢𝘷𝘪𝘳 ] [ 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥, 𝘝𝘰𝘪𝘤𝘦𝘥 𝘣𝘺 𝘈𝘐 ] What have I missed? See my last year's post on passing off AI content as your own: https://lnkd.in/d9h5MaZK 𝘞𝘢𝘯𝘵 𝘵𝘰 𝘴𝘦𝘦 𝘈𝘓𝘓 𝘮𝘺 𝘧𝘶𝘵𝘶𝘳𝘦 𝘱𝘰𝘴𝘵𝘴? 𝘊𝘭𝘪𝘤𝘬 🔔 𝘰𝘯 𝘮𝘺 𝘱𝘳𝘰𝘧𝘪𝘭𝘦 𝘱𝘢𝘨𝘦. Want to receive my weekly news digest straight to your inbox every Thursday morning? https://lnkd.in/dij4qFi5 Trying to position your tech venture as a leader in its space? That's what I do! https://lnkd.in/dwQ5JMtm #AIethics #ResponsibleAI [ 100% 𝘏𝘶𝘮𝘢𝘯 𝘊𝘳𝘦𝘢𝘵𝘦𝘥 ]