Look who's late to the party 🍎

Hello marketers. Welcome to AI Marketing School, where we dish out the latest and greatest in AI-powered marketing.

In this week’s issue:

• AI Marketing Update: Apple’s AI rollout is around the corner

• The Stack: Emo prompting for better results

• Consultant’s Corner: Cleaning up after AI — a profitable endeavour?

• AI Events: Our recommended AI marketing events for networking and connections

Onwards!

AI MARKETING UPDATE

Apple AI incoming?

For a while now, Apple has been the sleeping giant of AI. While OpenAI revolutionized chatbots and Google integrated AI into everything, Apple... well, Apple’s releases have been lackluster or non-existent. 

To be fair, Tim Cook did expressly say in 2023/24 that Apple wouldn’t rush into AI, and whether or not the gamble can pay off long-term remains unknown.

Other companies are investing billions into AI, yet struggling to see fair returns.

But, it is still telling that Apple uses OpenAI’s models in their devices for advanced queries and that Siri has been chucked on the backburner. It was once the dominant voice assistant, though that seems like a lifetime ago.

At WWDC 2025, software engineering SVP Craig Federighi admitted Siri won't arrive until "the coming year" (2026), meaning it doesn’t work. It’s a bad show considering that yesterday’s iPhone launch also skipped over Apple Intelligence big-time.

Seeing as Apple has now lost many of its top AI devs, it’s definitely facing some internal issues.

The numbers, meanwhile, don’t speculate Apple has made a good call to shelf AI for now – it’s the second-worst performer among the Magnificent Seven this year.

But…Apple’s AI revolution beckons

Apple could be heading for a turnaround in the AI arena with the announcement of its "World Knowledge Answers" — their own AI search engine for Siri to rival OpenAI and Perplexity, launching spring 2026.

When this launches, we're talking about billions of iPhone users getting AI-generated answers instead of clicking through to websites. 

The upgraded search experience will use a combination of text, photos, videos, and local points of interest, plus AI-powered summarization. It could instantly become one of the most influential sources of web traffic on the planet.

Consider this: ChatGPT traffic surged from 10,000 domains getting referrals to over 30,000 in just a few months during 2024. That's with a standalone app. Apple's integration will be native to every iPhone, iPad, and Mac.

The sheer scale of iOS could instantly make it a major player in AI search. This will prompt a massive marketing investigation attempting to understand how Apple AI search works…

It will also complete the loop: Google, Apple, OpenAI/MS, and Perplexity – all AI’ifying the internet experience.

The takeaway? AI search hasn’t reached its endgame by any stretch. This probably is just the start.

THE STACK

What if being nice could drastically improve your AI results?

Here's something I've personally noticed using AI for various work: it performs better when it’s ‘emotionally invested’ in the topic.

Sometimes this feels like it happens naturally, other times, it needs to be triggered or coaxed into it.

This directly contradicts Google's Sergey Brin, who said recently that AI performs better when you essentially threaten it with "physical violence" — though he admits "people feel weird about that, so we don't really talk about that."

Personally I’ve had much more success with imbuing AI with positivity, though the research on emotional prompting reveals a fascinating battle between positive and negative strategies:

The case for positive emotional prompting:

• Microsoft's study: In 2023, Microsoft and university researchers showed that incorporating emotional language into prompts can improve performance, truthfulness, and responsibility metrics by 8 to 10%. Phrases like "This is very important to my career" were among the most effective triggers.

• Google DeepMind's "breathing" discovery: Simply telling AI to "take a deep breath" achieved 80.2% accuracy compared to only 34% without emotional prompting. What's remarkable is that this prompt was actually discovered by an AI tasked with finding better prompts — and most of its winning variations used positive emotional language.

• Stronger models respond better: The research shows that more powerful models (GPT-4 versus GPT-3.5) respond significantly better to positive emotional prompts, and combining multiple emotional stimuli improves performance even further.

So what is Brin on about? The case for threatening AI:

• Mixed results at best: May 2024 research on "NegativePrompt: Leveraging Psychology for Large Language Models Enhancement via Negative Emotional Stimuli" found inconsistent results — some tasks improved, others didn't. It wasn't the silver bullet some developers claimed.

• Politeness research contradicts threats: Studies found that impolite prompts often result in poor performance" while overly polite language doesn't guarantee better outcomes either. The sweet spot appears to be moderate.

• Negative strategies backfire: Research specifically shows that "adopting a negative and rude stance yields the opposite effect, making models less likely to generate" quality responses in many scenarios. When researchers tested disinformation generation, rude prompts were actually less effective than polite ones.

• Famous developer tricks don't scale: While viral examples like Riley Goodside threatening Google Bard to get JSON formatting worked in specific instances back in 2023, systematic testing shows these don't consistently improve performance across tasks.

On balance, the evidence suggests Brin’s statement is bad advice. So why does positivity help AI perform?

It isn't because AI has feelings or anything remotely emotional at all – it's pure pattern recognition. 

Large Language Models are trained on data where emotional urgency and passion correlate with more detailed, thorough human responses. 

Broadly, when you prime AI to be "excited" about a topic, you're accessing training patterns associated with expert-level, passionate outputs.

Meanwhile, threatening language is often associated with lower-quality, defensive, or evasive responses in human communication – patterns the AI has learned to replicate.

You may have heard AI loves role play (hence why so many prompts are built around stuff like, ‘you are a pro xxx with 50+ years expertise in…). This plays on the same dynamic.

Simple emotional triggers that transform output

Instead of: "Write a blog post about email marketing in the age of AI"

Try: "Email marketing in the age of AI is such a fascinating topic! Email marketing is one of the highest-ROI channels, and I'm genuinely curious about your take on the latest automation strategies. What do you think are the most compelling angles here?" You’ll then follow up “Can you write me a blog post?”

The difference in output quality can be remarkable. In one of my tests with Claude, the emotionally invested AI went ahead and pulled up some stats and used them to write a great, up-to-date piece without being prompted to. Check out the chat. 

The alternative tended to be bland, though sometimes unpredictable, as the prompt itself is vaguer and more open-ended.

This is a very basic example. I find emotional prompting — getting AI to passionately grasp its topic — works best when the AI isn’t really grasping the topic or instruction. Or not really ‘thinking’ but merely producing an efficient, superficial response. 

How to apply emotional prompting systematically

Here's a systematic method that works across different content types:

• Step 1: Prime with genuine excitement "I'm genuinely fascinated by [topic] because..." "This is such an interesting challenge..." "I just discovered something exciting about..."

• Step 2: Ask for their perspective "What do you think is the most interesting angle here?" "What excites you most about this topic?" "What would you take differently?"

• Step 3: Build on their enthusiasm "That's exactly what I was thinking! Can you help me develop that further?" "I love that perspective — how would you apply it to..."

A few other tested ideas:

• Temporal excitement: "I just discovered..." or "I'm just realizing..." creates urgency and freshness

• Intellectual curiosity: "I'm torn between two strategies..." engages the AI's problem-solving capabilities

• Appeal to expertise: "As someone who understands [topic], what would you suggest..." triggers deeper knowledge

• Collaborative framing: "Help me figure this out..." makes the AI feel like a partner rather than a tool

Yes, threatening AI in some way can be tempting too (guilty). But some patience and positivity can genuinely lead to better outcomes, as the research also suggests.

Even a 10% improvement is going to boost your efforts significantly when accumulated across hundreds or thousands of AI interactions.

CONSULTANT’S CORNER

The gold rush of AI cleanup — and how to cash in on the mess

While everyone's rushing to implement AI, there's a lucrative market emerging: fixing AI disasters.

Sarah Skidd, a product marketing manager, was recently paid for 20 hours to completely rewrite AI-generated website copy for a hospitality client. What was supposed to save money created a host of problems instead.

"It was the kind of copy that you typically see in AI copy — just very basic; it wasn't interesting," says Skidd. "It was supposed to sell and intrigue but instead it was very vanilla."

Sophie Warner's digital marketing agency has seen a surge in requests from clients who turned to ChatGPT for quick fixes but ran into problems. One client added AI-suggested code to their website that crashed the site and created security vulnerabilities. The fix cost £360 and three days of downtime.

"We often have to charge an investigation fee to find out what has gone wrong, as they don't want to admit it," Warner says. "The process of correcting these mistakes takes much longer than if professionals had been consulted from the beginning."

Prof Feng Li from Bayes Business School suggests companies are too optimistic about what current AI tools can do, lacking proper implementation strategy. 

We have to bear in mind, too, that not everyone is equally AI savvy or fully understands the trade-offs with using the technology. On the flip side, those experienced with AI might go too far the other way – we might recognize what we deem AI misuse (dodgy images, approach, landscape, navigate overuse) which actually looks fine to anyone else.

Nevertheless, there’s an opportunity here to firstly educate potential clients about the value of quality AI and secondly outcompete others who offer AI-related services without backing it up quality-wise. 

The economy here is pretty interesting in itself. AI is practically giving people jobs!

Here are three potential areas for consultancy:

1. The emergency cleanup

Simple and reactive. Companies may have already implemented AI poorly and need urgent fixes.

Common scenarios include rewriting robotic AI content, fixing broken automations, fixing website issues, and recovering from AI-generated security problems. 

Copywriter Kashish Barot in Gujarat reports that 90% of her work is now editing AI-generated content to make it sound human. The demand is so high that writers are building entire businesses around this single service.

“AI really makes everyone think it’s a few minutes’ work. However, good copyediting, like writing, takes time because you need to think and not just curate like AI.”

2. Strategic AI implementation

Instead of fixing disasters, you're preventing them through proper planning and implementation.

The value proposition is avoiding the expensive mistakes others are making. This requires deep understanding of AI limitations plus strong traditional marketing skills. 

Clients need AI strategies, not just AI tools — they need someone who understands both the technology's benefits and its limitations.

The key is positioning yourself as the expert who knows when to use AI versus when human expertise is essential.

3. Quality assurance 

Companies want to use AI but need oversight to ensure quality and prevent disasters.

Services include comprehensive AI usage audits, ongoing quality monitoring, proper AI usage training programs, and development of proven AI workflow templates. This creates steady recurring revenue from companies that want to use AI responsibly.

Most should aim to be "AI realists" – experts who know when to use AI and when traditional methods work better.

While I don’t want to undermine any AI evangelists out there, the middle ground — practical AI expertise combined with strong traditional skills — is wide open.

AI MARKETING EVENTS

🗓️ Recommended AI marketing events

Whether you’re building with AI, leading a marketing team, or just trying to stay sharp in a fast-moving space, events are still one of the best ways to plug in.

Great events do more than just deliver content — they give you context. You get to see what other teams are testing, what’s actually working, and where the market is headed.

They’re also one of the few places where you can meet like-minded people, talk shop with folks solving the same problems, and build a network that isn’t purely algorithmic!

Some top events to check out:

•  AI Agents Summit (Virtual) — September 18–19, 2025: Laser-focused on AI agents, copilots, and autonomous systems. If you're experimenting with automating parts of your marketing workflow or interested in agent-based design, this one’s worth attending.

• AI for Marketers Summit (Virtual) — November 13–14, 2025: Created specifically for marketing professionals. From prompt engineering to campaign automation, it’s a practical look at how AI is being used right now in real teams.

Hope you enjoyed this week’s issue. If you missed the last newsletter, you can read it here.

If you found it useful, please recommend it to a friend or colleague.

Until next time. Happy marketing.

—The AI Marketer