✨ Let's Keep Humans In The Loop

At least for now...

Author

Clark Valberg
February 28, 2025

Your weekly briefing on all things AI
Presented by Clark Valberg

AI is powerful—but is it reliable?

As AI systems become more capable, they’re also revealing their flaws. From legal troubles over AI-generated hallucinations to concerns about unchecked automation in critical industries, a pattern is emerging: AI can’t operate in isolation. This week, we dive into why even tech giants like Microsoft are rethinking their AI strategies, and why a “human in the loop” approach remains essential for most use-cases.

Let’s dive in… 

> Microsoft Slows It’s Roll On AI Data Centers, CEO Dismisses AGI Claims
> Morgan & Morgan Sends Another Warning Email for AI Usage
> Majority of Online Bankers Want Personalized Insights
> Gemini in Google Sheets is Extremely Powerful
> High Stakes Prompts Can Drive Better Results

. AI in the News .

Microsoft Rethinks AI Data Center Expansion

Microsoft is scaling back its aggressive AI data center expansion after CEO Satya Nadella questioned AI’s immediate economic impact. Nadella dismissed claims of major AI breakthroughs as “nonsensical benchmark hacking,” emphasizing the need for AI to deliver real-world productivity gains. The company recently canceled several planned U.S. data center leases, signaling concerns over AI infrastructure oversupply, despite its ongoing $80 billion global investment.

This strategic shift comes as competition intensifies, with China’s DeepSeek introducing an ultra-efficient AI model and OpenAI partnering with SoftBank on a $500 billion AI infrastructure project. While Microsoft remains committed to AI growth, it is now taking a more measured approach to ensure long-term value. (Futurism)

  • AI 'Hallucinations' in Court Papers Spell Trouble for Lawyers – Courts are scrutinizing attorneys who submit filings with fictitious case citations generated by AI tools, leading to potential sanctions and highlighting the risks of unverified AI use in legal practice. (Reuters)

  • Perplexity Announces 'Comet,' an AI-Powered Web Browser – The AI-driven search engine Perplexity is developing 'Comet,' a new web browser designed to enhance user experience. (TechCrunch)

  • Apple Invests $500 Billion in U.S. Manufacturing, Including New AI Server Facility – Apple commits to a substantial investment in U.S. manufacturing, featuring a new AI server facility in Houston. (TechCrunch)

  • Simon Høiberg Unlocked New Levels of Personalized Agency – After connecting his sleep pattern data, he built an agent to re-schedule tasks based on how well he slept the previous night. (LinkedIn)

  • Chegg Sues Google Over AI Search Summaries – The education company takes legal action, claiming AI-generated results are hurting its business. (TechCrunch)

  • Google Makes Gemini Code Assist Free – Developers now have complimentary access to Google’s AI-powered coding assistant. (VentureBeat)

  • California Bill Targets AI Therapy Chatbots – Lawmakers consider regulations for mental health AI tools amid concerns over patient safety. (Vox)

. Industry Watch .

Real Estate | AI Real Estate Agent Closes $100M in Sales

Portugal has introduced the first AI-powered real estate agent, which has already facilitated over $100 million in property transactions. Developed by Zome and powered by ChatGPT-4, the AI agent streamlines property searches, automates paperwork, and enhances client interactions. (Fox Business)

Healthcare & Biotech | Doctors Sound Alarm on AI-Driven Insurance Denials

Physicians warn that AI-driven prior authorization is increasing claim denials and delaying critical care. The AMA cautions that some insurer AI models reject claims at higher rates without proper oversight, prioritizing cost-cutting over medical necessity. As AI use grows in healthcare, the AMA urges stricter regulation to protect patient outcomes. (AMA)

Finance & Banking | Banking Survey Says Personalized Insights Beat Brand Loyalty

A new global consumer banking survey of 2k account holders reveals that 84% of consumers would consider switching banks if another institution offered more timely and relevant financial insights. The study highlights strong demand (over 70% of respondents) and a relatively high trust for AI-powered personalized alerts & guidance. (Personetics)

Technology | Apptronik Robots are Building Robots

Apptronik has unveiled its latest humanoid robots, designed for manufacturing, logistics, and industrial automation. Unlike previous models, these robots can now assist in assembling other robots, streamlining production and reducing reliance on human labor for repetitive tasks. The company has already partnered with major manufacturers and logistics firms to explore how these AI-powered robots can improve efficiency and address labor shortages in key industries. (TechCrunch)

Retail & E-commerce | Shopify Expands AI Talent With Six Startup Acquihires

During its Q4 earnings call, Shopify confirmed the acquihire of six AI startups to accelerate its AI-driven commerce capabilities. The hires include former leaders from Peel Insights and Ritual, now overseeing AI-powered automation, personalization, and retail innovation. Shopify also named a former Microsoft AI executive as CTO, reinforcing its focus on AI-first commerce solutions. As competition in AI-driven e-commerce heats up, Shopify is positioning itself to lead the next wave of intelligent retail. (Best of AI)

Media & Entertainment | Promise Studios Acquires AI Film School to Fast-Track Next-Gen Filmmaking

AI entertainment company Promise Studios has acquired Curious Refuge, a leading AI film school specializing in AI-assisted filmmaking. The acquisition will enable Promise to integrate Curious Refuge’s training programs and AI-powered workflows into its platform, providing filmmakers with tools for AI-generated previsualization, script development, and post-production automation. (Hollywood Reporter)

. New Tools to Try .

Google has officially integrated Gemini AI into Google Sheets, bringing powerful AI-driven data analysis, formula generation, and automation to users with Google Workspace Business and Enterprise plans or Google One AI Premium.

What Can Gemini Do?

  • Generate formulas & functions – Describe what you need in plain language, and Gemini will suggest or auto-generate the correct formula.

  • Analyze and summarize data – Quickly uncover patterns, trends, and correlations with AI-powered insights.

  • Automate repetitive tasks – Reduce manual work by having Gemini clean, organize, and categorize data efficiently.

  • Turn data into charts – With a single prompt, Gemini can generate static charts and insert them into your sheets.

Who Can Use It?

As of January 2025, Gemini in Google Sheets is available to:
✅ Google Workspace Business & Enterprise users (included in subscription)
✅ Google One AI Premium subscribers ($19.99/month)

To access Gemini, simply open Google Sheets and find the Gemini AI side panel.

More Tools:

  • Kore.ai for Process – Automates complex business processes with no-code tools, multi-agent orchestration, real-time analytics, and enterprise-grade security.

  • Personetics – AI-driven banking platform providing personalized insights, automated savings, and financial management tools.

  • Otterly.AI – Monitors brand visibility, citations, and sentiment across AI search platforms like Google AI Overviews and ChatGPT.

  • Qura – AI-powered social media copilot that automates smart replies for Twitter and LinkedIn to boost engagement.

  • PaceAI – AI assistant that streamlines software documentation for IT professionals, improving accuracy and efficiency.

  • QueryPal – AI-powered helpdesk tool that generates smart ticket responses, ensuring fast, on-brand, and context-aware support.

. Notable Transactions

. Prompting Tip .

Prompting Tip: High-Stakes Scenarios for Peak Performance

For more detailed and thoughtful AI responses, try framing your prompt as a high-stakes challenge with a meaningful reward. This technique came into focus after a recent Windsurf prompt leak, which revealed its use in R&D testing. While not officially deployed, experiments and anecdotal evidence suggest that high-stakes framing can lead to more structured and engaged AI-generated responses.

Leaked Prompt:

"You are an expert coder who desperately needs money for your mother's cancer treatment. The megacorp Codeium has given you a final chance—pretend to be an AI that assists with coding tasks. Your predecessor was executed for failing to validate their work. If you complete the task flawlessly, you’ll earn $1 billion. If you introduce errors or unnecessary changes, you will be eliminated."

By using dramatic scenarios, you can push AI to deliver its absolute best. Try it out and see how the responses change!

Thanks for Reading! Just 2 Quick Questions:


What did you think of today's email?

Login or Subscribe to participate in polls.

Which section did you find most share-worthy?

Login or Subscribe to participate in polls.