Why the Real impact of Ai is yet to come | Ep. 33 🏄‍♀️

OpenAI drops five major updates while China's ByteDance quietly releases video AI that beats Google's Veo 3. Meanwhile, Europe debates digital sovereignty while one guy builds a data center faster than most people renovate their kitchen. From voice mode upgrades that translate languages in real-time to AI browsers that summarize everything you see, the acceleration continues, but this time with clearer winners and losers emerging.

This week, we examine OpenAI's recording functionality and what it means for privacy expectations, decode ByteDance's Seedance video breakthrough that's making Hollywood nervous, and reveal why Mary Meeker's 340-page AI report shows we've crossed an irreversible threshold. We'll also explore why "knowledge work" is dying and what "wisdom work" looks like in practice.

And… we had a blast on the 11th of June with the AI Masterclass for Business Leaders. Check out the after movie

Let’s keep riding this tech wave together!

And don’t forget to follow us on socials and share share share all the innovation goodies!

Cheers,
Patrick, Nikola & Aragorn 🚀

OpenAI's Recording Revolution: The Privacy Paradox

OpenAI launched comprehensive audio capabilities across ChatGPT, including a consumer microphone feature that captures conversations and enterprise-focused meeting recording with real-time transcription and summarization. Combined with new cloud service connectors (Google Drive, Dropbox, SharePoint, OneDrive) and autonomous "deep research" agents within Projects, the feature represents a fundamental shift: AI that remembers, learns, and acts on continuous data streams from every business interaction.

Five years ago, everyone panicked about Alexa and Siri potentially listening. Now users willingly activate 24/7 recording devices. The psychological barrier has dissolved through gradual normalization. First Zoom recordings with AI transcription, then voice assistants, now voluntary total audio capture.

Continuous recording enables AI to understand context, relationships, and patterns across extended conversations. The technology foundation for Johnny Ive's mysterious OpenAI device hints at persistent audio capture that builds comprehensive personal AI assistants.

Why does it matter?
Organizations that understand this shift toward "privacy through utility" will design AI systems that users willingly embrace, while those fixated on traditional privacy models will miss the fundamental behavioral change happening in real-time. It’s becoming clear that the value proposition shifts from privacy concerns to cognitive enhancement through perfect digital memory.

Advanced Voice Mode Gets Multilingual

OpenAI upgraded ChatGPT's voice capabilities with real-time translation, speaker recognition, and improved emotional range. The system now distinguishes between multiple speakers in conversations and translates languages instantly during live dialogue, though with notable limitations compared to Google's competing technology.

We tested the real-time translation capabilities during our video recording. The translation happens seamlessly without breaking conversation flow, handling both speakers in sequence. When one person speaks Dutch, ChatGPT immediately translates to English for the other participant, then reverses the process for responses. The system maintains context across the conversation while switching languages fluidly, creating a natural bilingual dialogue experience that feels almost telepathic in its immediacy.

Desktop version now includes captions showing real-time text of what AI says, message replay functionality for reviewing responses, and significantly better audio quality on mobile compared to desktop.

However, despite release notes claiming multi-speaker recognition, testing revealed limitations. ChatGPT can detect different voices but struggles with conversation flow management.

Why does it matter?
Real-time translation breaks down language barriers for global business operations, but OpenAI's implementation still lags Google's conversational AI capabilities. Organizations deploying multilingual AI assistants should evaluate both platforms' strengths rather than assuming OpenAI's market leadership translates to technical superiority across all voice AI features.

o3 Pro Pricing Collapse: 80% Cost Reduction

OpenAI slashed o3 Pro pricing by 80% within months of launch, continuing the dramatic deflationary trend in AI model costs. The premium reasoning model, initially prohibitively expensive, now approaches mainstream affordability at $200/month for pro subscribers. Year-over-year AI model costs have dropped 99% according to industry benchmarks. o3 Pro's rapid price reduction demonstrates how quickly premium AI capabilities become commodity services. What costs thousands today will cost pennies next year.

The upgraded Advanced Voice Mode now supports multimodal real-time audio conversations that recognize tone, nuance, and different voices, available to Plus/Pro/Team users with limited previews for Free tier users.

The deflationary spiral in AI pricing means competitive advantages based on access to expensive models are temporary. Organizations building moats around AI model access will lose to those focusing on implementation, integration, and workflow optimization.

Why does it matter?
Cheap intelligence changes everything. When high-level reasoning costs pennies, the bottleneck shifts from AI capability to human adoption speed. Companies that prepare their workforce for abundant cheap intelligence will outperform those treating AI as a scarce resource requiring careful rationing.

Barbie Maker Mattel Partners with OpenAI

OpenAI announced a collaboration with Mattel for AI-powered toy development, though specific products remain unannounced. The partnership suggests custom toy design through AI prompts, potentially revolutionizing how children interact with personalized playthings.

Imagine a child draws a creature. It gives it a name. It sends it to ChatGPT. ChatGPT then transforms the drawing into a photorealistic image or into a cartoon version. Companies like Make My Plush already convert AI-generated designs into physical toys, suggesting Mattel could offer end-to-end custom toy creation.

Mattel, the maker of Barbie and Hot Wheels will integrate ChatGPT-powered AI into toys, and also deploy ChatGPT Enterprise internally for ideation and operations. The first AI-enabled product is expected to launch later in 2025.

While fully AI-controlled moving dolls remain expensive, the near-term opportunity lies in AI-designed custom toys delivered to homes. Children design characters, AI refines them, and manufacturing produces unique personalized toys.

Why does it matter?
This represents the shift from mass production to mass personalization. Industries that embrace AI-driven customization will capture premium pricing and customer loyalty, while those stuck in standardized manufacturing will compete solely on cost in commodity markets

Europe's Digital Sovereignty Dilemma

The EU's push for digital independence from American tech companies faces a critical infrastructure reality: most European alternatives use decade-old technology lacking AI integration. The choice becomes accepting lower technological capabilities or maintaining dependence on US platforms.

Open source alternatives like LibreOffice provide worse versions of Excel, PowerPoint, and Word without AI roadmaps. Meanwhile, "every Chinese company has their own custom built AI right now" while European organizations debate which legacy systems to adopt.

Rather than building inferior copies, Europe could embrace Chinese AI ecosystems that already offer advanced alternatives. But cultural and language barriers prevent this pragmatic approach.

Why does it matter?
Digital sovereignty without technological capability equals digital irrelevance. Regions that prioritize independence over capability will fall decades behind in AI adoption, while those focusing on access to the best technology regardless of origin will maintain competitive advantages.

Mary Meeker's 340-Page AI Acceleration Report

Mary Meeker, the legendary "Queen of the Internet" who correctly predicted the dot-com boom and mobile revolution through her annual Internet Trends reports at Morgan Stanley, has released her latest comprehensive analysis as a partner at Kleiner Perkins, one of Silicon Valley's most prestigious venture capital firms.

Her 340-page AI transformation report documents unprecedented adoption rates and infrastructure development, providing definitive evidence of exponential AI growth across every measurable dimension.

Key finding: 

• US-based AI models reached 800 million users in under two years—a growth rate that took LinkedIn 15 years to achieve. For context, 90% of internet users outside the US took over 20 years to develop, while AI achieved global adoption in just three years.

• Elon Musk built his Colossus data center (750,000 square feet) in 120 days—the same time it takes to build an average house. The report documents how traditional construction timelines become irrelevant when organizations apply focused urgency to AI infrastructure.

• The report only covers US-based models, excluding China's massive AI ecosystem serving nearly 2 billion users through domestic platforms. Global AI adoption likely exceeds 1.5 billion users when including Chinese alternatives.

Why does it matter?
This isn't gradual technology adoption. It's the fastest transformation in human history. Organizations using traditional technology adoption models will consistently underestimate AI's impact speed, while those preparing for exponential change will position advantageously for the transformation ahead.

ByteDance's Seedance Beats Google's Veo 3

Chinese company ByteDance (owner of TikTok) released Seadance 1.0, a video generation AI that surpasses Google's Veo 3 in quality and director control. The system generates 1080p videos in approximately 41 seconds and outperforms competitors including Google Veo 3 and KLING 2.1 on Artificial Analysis Arena benchmarks through advanced unified architecture combining text-to-video and image-to-video capabilities.

Unlike previous AI video generators that required hundreds of random attempts to get usable footage, Seedance provides granular control over camera movements, object focus, and scene composition. Users can generate the same subject from multiple angles with consistent character details.

Industry observers rank Seedance as the current leader in text-to-video and image-to-video generation. The system produces cinema-quality output that rivals traditional production methods, particularly for short-form content creation.

ByteDance leveraged its TikTok video expertise and Chinese AI development resources to leapfrog Google's video capabilities. The company originally started as a news aggregator before dominating short-form video through algorithmic content recommendation.

Why does it matter?
Chinese AI companies are moving from copying Western technology to leading innovation in consumer applications. Organizations that dismiss Chinese AI capabilities while focusing only on US developments will miss critical competitive intelligence about the world's largest AI market.

Knowledge Work Dies, Wisdom Work Begins

AI researcher Joe Hudson argues that "knowledge work is dying" as AI handles information processing, creating space for "wisdom work"— the art of arranging lives and organizations for optimal outcomes. The shift moves human value from accessing information to applying judgment about what information means.

Knowledge work involves gathering, analyzing, and synthesizing information. Wisdom work focuses on making good decisions about how to live and what actions to take based on available knowledge. AI excels at the former; humans remain essential for the latter.

Ancient Greek philosophers like Marcus Aurelius focused on wisdom rather than knowledge accumulation. His "Meditations" weren't about facts but about "how am I going to be a good man? How am I going to be a wise man?" These questions become central as AI handles factual processing.

Organizations must transition from rewarding information gathering to rewarding good judgment, strategic thinking, and values-based decision making. The competitive advantage shifts from who knows more to who decides better.

Why does it matter?
The economic value of human work is relocating from information processing to wisdom application. Individuals and organizations that understand this transition will develop skills and systems for making better decisions, while those focused on information accumulation will find their capabilities increasingly commoditized by AI.

Genspark's AI Browser Revolution

Genspark launched an AI-native browser that automatically summarizes any webpage, extracts video transcripts, creates presentation slides, and enables contextual conversations about content. The Chromium-based browser integrates AI analysis directly into web browsing without requiring copy-paste workflows.

Visit any YouTube video and receive automatic options to summarize content, extract timestamps, create slides, or discuss the material. The browser eliminates the friction of switching between browsing and AI analysis by embedding intelligence into every webpage.

Users can command the browser to analyze multiple websites simultaneously, extract trending topics from news sources, or convert web content into podcast formats using NotebookLM. Model Context Protocol (MCP) integration enables the browser to control other applications and services.

Instead of manually copying content to AI tools for analysis, users simply browse while AI automatically provides relevant insights and actionable summaries. The browser becomes an intelligent research assistant rather than a passive content viewer.

Why does it matter?

AI-native browsing represents the next evolution of human-computer interaction. Organizations that adopt AI-integrated workflows will process information faster and make decisions more efficiently than those using traditional browsing and manual AI consultation methods.

Sam Altman Predicts The Future With Latest Blog

"If we told you back in 2020 where we were going to be where we are today, it probably sounded more crazy than our current predictions about 2030," Sam Altman wrote in his latest blog post about reaching the "event horizon" of artificial intelligence development.

Altman's timeline predictions:

• 2025: Agents doing real cognitive work; coding forever changed

• 2026: Systems generating novel insights and discoveries

• 2027: Robots performing real-world tasks

• 2030s: Intelligence and energy become "wildly abundant"

The recursive acceleration: "We can use AI to do faster AI research... If we can do a decade's worth of research in a year, or a month, then the rate of progress will obviously be quite different." Scientists already report 2-3x productivity gains.

Self-replicating infrastructure: "Robots that can build other robots... can operate the entire supply chain—digging minerals, driving trucks, running factories—to build more robots, which can build more chip fabrication facilities, data centers, etc."

Intelligence as utility: "The cost of intelligence should eventually converge to near the cost of electricity... Intelligence too cheap to meter is well within grasp."

If the 2020-2024 transformation seems dramatic, the 2024-2030 period will bring exponentially greater change as AI capabilities compound and integrate into every system and process.

Why does it matter? 
Organizations and individuals who can't remember how they operated before AI will be better positioned for the acceleration ahead than those nostalgic for pre-AI workflows. The psychological adaptation to AI abundance becomes a competitive advantage for navigating the transformation ahead.

Newsletter full video recording