Friday the 28th of November

G’day folks,

It’s Thursday the 28th of November and researchers in Sweden have developed AI that can detect dementia with 97% accuracy using simple brain scans.

At the same time, an MIT study is revealing that AI can already replace nearly 12% of the American workforce, and Robinhood’s CEO just turned his side project into a $1.45 billion AI unicorn focused on mathematical superintelligence.

Let’s get into it.

This newsletter is powered by:

• Letz - Create what you love

• Expo - Everything you need to build apps

• Firecrawl - The Web Data API for AI

• Anything - Turn your words into mobile apps, sites, tools, and products

• Browserbase - A web browser for your AI

• 9x - Upskill your teams in AI

• AI Workforce Disruption: A new MIT study using the “Iceberg Index” reveals that artificial intelligence can already replace 11.7% of U.S. jobs, equivalent to $1.2 trillion in wages, with exposure far beyond tech roles into finance, healthcare, and administrative functions.

• Dementia Detection Breakthrough: Swedish researchers at Örebro University have developed AI models that analyze EEG brain signals to detect dementia with over 97% accuracy, potentially enabling early diagnosis through simple, low-cost testing that could eventually be conducted at home.

• Regulatory Showdown Brewing: The race to regulate AI has sparked a federal versus state battle, with 38 states passing over 100 AI laws this year while the White House drafts proposals to preempt state regulations and establish federal supremacy.

• Math AI Hits Unicorn Status: Harmonic, the AI startup co-founded by Robinhood CEO Vlad Tenev, has raised $120 million at a $1.45 billion valuation, focusing on building “mathematical superintelligence” with error-free reasoning capabilities.

The Massachusetts Institute of Technology has released a groundbreaking study showing that artificial intelligence can already replace 11.7% of the U.S. labor market—about $1.2 trillion in wages across finance, healthcare, and professional services. The findings challenge assumptions that AI disruption will remain confined to coastal tech hubs.

The details:

• The study used a labor simulation tool called the “Iceberg Index,” created by MIT and Oak Ridge National Laboratory, which models how 151 million U.S. workers would be affected by AI adoption.

• The visible “tip of the iceberg”—layoffs in tech and IT—represents just 2.2% of total wage exposure ($211 billion), while the bulk lies in routine functions in HR, logistics, finance, and office administration.

• The index maps over 32,000 skills across 923 occupations in 3,000 counties, measuring where current AI systems can already perform those skills without future speculation.

• Tennessee, North Carolina, and Utah have already begun using the platform to build policy scenarios and workforce training programs.

Why it matters: This isn’t a prediction about future automation—it’s a snapshot of what today’s AI can already do. The findings give policymakers a county-by-county roadmap of AI exposure, enabling targeted reskilling investments before disruption arrives. CNBC

Researchers at Örebro University in Sweden have developed two revolutionary AI models that can analyse electroencephalogram (EEG) brain signals to detect dementia—including Alzheimer’s disease—with over 97% accuracy. The breakthrough could transform early diagnosis by making testing simpler, cheaper, and more accessible.

The details:

• The first model combines temporal convolutional networks with LSTM networks to analyse EEG signals, distinguishing between Alzheimer’s, frontotemporal dementia, and healthy individuals with over 80% accuracy.

• The second model is remarkably efficient—under one megabyte in size—and uses federated learning to protect patient privacy while achieving 97% accuracy.

• The AI identifies patterns in different brain wave frequencies (alpha, beta, gamma) and uses explainable AI technology to show doctors exactly how it reaches its conclusions—no more “black box” diagnoses.

• EEG testing is already simple and inexpensive, and combined with AI that can run on portable devices, this opens the door to widespread use in primary care and potentially future home testing.

Why it matters: Early dementia diagnosis is crucial for slowing disease progression and improving quality of life. By making detection faster, cheaper, and privacy-safe, this AI could ease the burden on patients, caregivers, and healthcare systems worldwide. The researchers plan to expand to larger datasets and include other dementia types like vascular dementia and Lewy body dementia. News Medical

The race to regulate artificial intelligence has ignited a high-stakes showdown between federal and state governments, with the White House preparing executive orders to block state AI laws while 38 states have already passed over 100 AI regulations this year.

The details:

• As of November 2025, states have adopted more than 100 AI-related laws, mainly targeting deepfakes, transparency requirements, and disclosure obligations.

• The White House has drafted a proposal to direct federal agencies to identify “burdensome” state AI regulations and pressure states not to enact them, arguing for unified federal standards.

• Congress is gearing up for a year-end fight over a proposed moratorium on state AI rules, with tech companies pushing for federal preemption while state lawmakers defend their role as “laboratories of democracy.”

• The tension reflects a fundamental question: Should innovation-friendly federal oversight prevail, or should states maintain flexibility to address local concerns about AI safety and ethics?

Why it matters: This regulatory battle will determine whether AI companies face a patchwork of 50 different state laws or a single federal framework. The outcome will shape everything from innovation speed to consumer protection, with implications for startups and tech giants alike. States argue they’re filling a federal vacuum, while the White House warns that conflicting state rules could stifle AI development. TechCrunch

• Canada Invests in AI Infrastructure: The Canadian government committed $42.5 million to expand AI compute infrastructure at the University of Toronto, enabling researchers across the country to train AI models with billions of parameters and reducing dependence on foreign providers like Nvidia and Microsoft.

• Majestic Labs Emerges from Stealth: Led by ex-Google and Meta executives, AI infrastructure startup Majestic Labs raised $100 million to build next-generation servers that solve the “memory wall” problem, promising 1000x greater GPU memory capacity for memory-intensive AI workloads.

• Robinhood CEO’s AI Startup Soars: Harmonic, co-founded by Robinhood CEO Vlad Tenev, reached a $1.45 billion valuation after raising $120 million in Series C funding to develop AI focused on mathematical reasoning and error-free computation—positioning itself as a specialist in “mathematical superintelligence.”

• AI Adoption Rates Flatten: New data suggests AI adoption rates are starting to plateau, raising questions about whether the technology has reached an early saturation point or if companies are pausing to consolidate existing implementations before the next wave of expansion.

That’s all for today,

See you tomorrow. Tom