Today, we’re bringing you the latest in AI-powered marketing and business strategies. Here’s what’s inside:

🚨 AI Top Story: AI can hallucinate even when it knows the right answer—here’s why it happens and how to prevent it.

📊 One Quick AI Hack: Get AI to fact-check AI—running the same query through another model can catch errors before they spread.

🎯 Killer Marketing Prompt: Virality isn’t luck—it’s a pattern. Use AI deep research to uncover what makes content in your industry take off.

🌟 Creator Spotlight: Valeriya Pilkevich shares 3 innovative marketing use cases for Perplexity’s DeepResearch Model.

AI’s Most Frustrating Habit: Lying with Confidence

Why AI Hallucinates—And What You Can Do About It

We’ve all heard about AI hallucinations—those moments when AI confidently spits out completely false information. Usually, this happens when the AI simply doesn’t have the right answer and decides to fill in the blanks. But here’s the wild part: AI can still hallucinate even when it actually does have the right answer.

That’s right. AI sometimes “misses the boat” and generates a made-up response despite knowing the correct information. Recent research shows that even when AI has the facts, it can still veer off course, delivering confabulated nonsense instead of what it already knows to be true.

Why? Because AI doesn’t "think" the way humans do—it generates responses based on probabilities, and sometimes, that process goes sideways. A lack of specificity in prompts can also cause AI to take creative liberties where it shouldn’t. Broad or vague instructions often lead to responses that prioritise coherence over factual accuracy.

For marketers using AI in content creation, chatbots, or customer service, this is a big deal. Imagine an AI-powered assistant giving customers incorrect product information when the correct details were sitting right there in its training data. Or an AI-generated report fabricating numbers even though the real data was available.

These “missed-the-boat” hallucinations can erode trust fast—especially when AI outputs are assumed to be reliable at first glance. That’s why specificity matters. Breaking down complex requests into smaller, well-defined prompts helps AI stay on track. Instead of a general request like “Analyze customer feedback”, a better approach would be: “Summarise common themes in customer reviews” and “Identify recurring complaints in product feedback” separately.

Another critical safeguard is allowing AI to acknowledge uncertainty rather than forcing it to fill in gaps. When AI is unsure, prompting it to explicitly state "I don't know" instead of generating an answer can prevent misinformation. This is especially important in customer support or data analysis, where false confidence can lead to serious business implications. Additionally, asking AI to cite specific sources when available—rather than generating broad claims—adds another layer of reliability.

Even with these precautions, AI-generated content should never be taken at face value. Verifying responses against known data sources and running multiple iterations of a prompt to compare consistency can highlight potential errors before they make their way into reports, articles, or customer interactions. AI doesn’t “think” like a human, but when tasked with synthesising multiple responses and combining the most reliable insights, it can approximate something closer to thoughtful analysis.

🎨 “WTF is happening?” – how AI is reshaping design - Seasoned designers are adopting AI tools at higher rates to enhance creativity & efficiency

🤖 Musk v Altman: What might really be behind failed bid for OpenAI - Elon Musk's recent $97.4 billion bid to acquire OpenAI has been rejected

📊 Solving the data crisis in generative AI: Tackling the LLM brain drain - Generative AI models are facing a data crisis as public data sources become restricted

🤝 AI and Humans: A Complex Partnership in the Workplace - A recent study reveals that while AI can augment human skills, closer human-AI collaboration may not always yield better results

🎯 The Artificially Intelligent Enterprise - The Ultimate AI Research Assistant

☕ AI Tangle - Elon's xAI Debuts its Long Overdue Grok 3 Together With Reasoning

Get AI to give a second opinion on your first AI’s opinion

Sounds counter intuitive, I know. . but since different models are trained on varying datasets, running the same query through another chatbot can help spot inconsistencies and highlight potential errors.

This is especially useful when dealing with data-heavy responses, summarising research, or generating content that requires a high degree of accuracy. If two AI models provide conflicting answers, that’s a clear signal to dig deeper.

To make this process more effective, refine your prompts to force AI to scrutinise its own output.

Instead of asking “Is this correct?”, try:

 For verifying factual accuracy:
"Cross-check this response with verified sources. If discrepancies exist, explain the differences and suggest the most reliable answer."

For checking numerical data or statistics:
"Validate these figures by referencing multiple datasets. If conflicting values appear, provide the most consistent range and identify possible reasons for variation."

For ensuring AI doesn’t fabricate information:
"Confirm whether all details in this response can be traced to known sources. If any part is speculative or unverifiable, flag it and suggest a clearer alternative."

This quick check can save time and prevent errors from making their way into reports, marketing materials, or customer interactions. Even when AI gets things right most of the time, a second opinion never hurts.

Valeriya Pilkevich - Perplexity just dropped its own "Deep Research," and it's already proving useful for marketers. Check out these three use cases!

Reverse-Engineer Virality for Your Industry in AI Deep Research

This week’s Killer Marketing Prompt is simple, but surprisingly effective.

Virality isn’t random—it follows patterns. This Deep Research prompt uncovers what makes content in your industry take off.

The Prompt:

Analyse viral marketing campaigns in [industry]. Identify recurring patterns, emotional triggers, content formats, and distribution strategies that contributed to high engagement and shareability. Break down key success factors, including audience psychology, timing, and platform-specific trends. Compare these insights to recent campaigns in [your brand/competitor] to highlight potential opportunities for replication or innovation.

People often think that there is a secret hack to going viral. The truth is, no matter what hashtags you use, no matter how many times you refresh the page - viral content comes down to one thing and one thing only. The content itself.

DeepSeek has replaced ChatGPT in so many people’s AI Stack.
Which LLM’s are you using most these days?
Let us know in our survey below!