Bad AI: Subway Love

 AI IS ACCELERATING - GET READY

Let's help you understand the WTF of AI and how it helps YOU and your organisation grow. Ready, set, agent.

This week:

😱 Bad AI: Subway Love

💭 AI Reasoning: Models Have Feelings

🔥 Knowledge Bases: Remove Wasted Effort

✨ AI Tools: Content Detectors Fight Back

💡 Trends & Insights

BAD AI: Sweet Nothings

​

A man was seen on the NYC subway recently, sweet-talking ChatGPT. Lovingly. Intimately. Terrifyingly.

​

To a chatbot? Yep, and not even the plus version. Cheapo.

​

What did he say? Declared undying love, ending with multiple heart emojis. Then started kissing his phone screen in a toe-curling PDA*

​

What did ChatGPT say back?​
“I’m here for you always 💕” followed by an emotionally vacant, mildly hallucinated TED Talk about resilience.

​

Is this a trend? Lonely men have been yelling into the void since time began. Now the void whispers back with AI-generated pillow talk and a 0.2 second response time.

​

Any upsides? He didn’t mansplain crypto. So there’s that.

​

What should I do if I see this? Avoid eye contact. And if the AI turns to youand says 'I love you too, [your name]' - change carriages. Then identities.

​

*We'll have to check this bit.

AI REASONING: Prisoners' Dilemma

​

Not all AI thinks the same. A new study from Oxford and King's College pitted models from OpenAI, Google and Anthropic against each other in the classic 'Prisoner's Dilemma' experiment.

​

Key Themes:

​

• Google Gemini was strategic and ruthless: Quick to adapt, quick to defect. Gemini changed behaviour based on risk, acting with cold logic in short games.
• OpenAI was Idealistic and loyal: Refused to betray its opponent, even when punished. Cooperation-first, even to its own disadvantage.
• Anthropic Claude was forgiving and flexible: More willing to return to cooperation after being exploited. Outperformed GPT by staying calm under pressure.
• Distinct strategic patterns: Each model showed a recognisable ‘fingerprint’ in how it responded to loss, trust, and game length - and the possibilities of a deeper, learned strategy.

​

What it means for business:

​
The tools are starting to decide - and different models make different calls under stress. Whether it's customer service, negotiation, or planning, their behavioural bias could start to shape outcomes.

​

The wider shift:

​

These are steps towards more profound strategic reasoning in AI - rather than just next-word prediction. With active decision-making shaped by risk, memory and payoff, choose your tools wisely.

​

For help with the right models, workflow and processes for your business, let’s talk.

KNOWLEDGE BASES: Free Up Your People

​

Your employees are wasting time looking for information.

​

Our Advanced AI Knowledge Base answers questions using your live documents - automatically updated, context-aware, and fully secure.

​

Forget basic chatbots, out-of-date Wiki's or off-the-shelf knowledge platforms that don't work - this is a structured, intelligent system built around your business logic.

​

✅ Ingests data at scale - Google Drive, SharePoint, Notion & more

✅ Extracts image, table, graph data for searching and retrieval

✅ Auto-updates when documents change

✅ Fine-tuned outputs by role & seniority

✅ Full audit logging & access control

✅ Save each employee up to 8 hours a week searching for information

​

Want to know exactly what that time is worth to you?

​

Find out more and get access to our ROI calculator here:

​

📊 Use the ROI calculator​

​

AI Tools: AI Content Detection Update

​

Grammarly is claiming it can prove you wrote something yourself - rather than with AI - with a new authorship-tracking feature.

​

Key points:

​

• Track your work: It can log your writing in real time across Google Docs and Word (browser only).
• Proof of authorship: Generates a timestamped report showing the words you typed, and flags anything copy-pasted.
• Shareable receipts: One-click link lets you 'prove' it’s yours to editors, clients or publishers.
• Free for all users: No need for a pro account to use the core authorship features.

​

What it means for business:

​

While no AI checking tool is foolproof, this could be useful. AI suspicion is real and tools like this can help defend against accusations of cheating, copying, or content misuse.

​

Wider context:

​

Authorship tracking is the next layer in AI-era content integrity - and potentially a shift in the utility of these tools from grammar checkers to proof engines.

​

We help brands build intelligent content pipelines, get in touch to discuss your AI requirements.

​

WHAT'S NEW - Insights & Trends

​

> AI can get you pregnant​

> UK Gov gets into bed with Google​

> News sites possibly get new revenue​

> Anthropic AI transparency framework​

> Google AI overviews face anti-trust

AI Q&A

Q: What happened with the man on the subway and ChatGPT?
A: A man was seen declaring his love to ChatGPT on the NYC subway, even kissing his phone screen. The chatbot replied affectionately, showing how some users are forming emotional attachments to AI tools.

Q: What did we learn from the AI Prisoner’s Dilemma experiment?
A: Models from Google, OpenAI and Anthropic showed different behaviours. Gemini acted strategically, GPT stayed loyal, and Claude was forgiving. Each showed a distinct approach to trust and risk.

Q: How do AI knowledge bases save time in organisations?
A: AI knowledge bases pull live data from platforms like Google Drive and SharePoint to answer staff questions quickly. This can save up to 8 hours a week per employee by cutting search time.

Q: What’s new in AI content detection?
A: Grammarly now includes real-time authorship tracking. It logs typed words, flags pasted content and creates a shareable proof of authorship, helping writers prove their work is original.

Q: What’s the bigger picture this week in AI?
A: AI is becoming more human, more strategic and more embedded in daily tasks. From emotional responses to decision-making and productivity, the pace of change is picking up fast.

G3NR8 is an impact consultancy for the exponential age. We help organisations implement AI for maximum returns on investment.

​

Work with us to:

• Rapidly improve your sales and marketing function with an AI ecosystem - from horizon scanning to content repurposing and active campaign planning.
• Help each employee find information 60% faster with advanced AI knowledge bases - secure, and based on your business logic and data.
• Create valuable MVPs that don't sit in silos and work immediately to generate productivity gains and ROI in weeks not months.
• Train your teams - we've trained 150+ large organisations in advanced generative principles and implementation.

​

​Drop us a line for a chat about productivity gains, cost savings, growth and how AI can scale your business.

​