In this episode of Stewart Squared, host Stewart Alsop III talks with his father, Stewart Alsop II, covering a wide range of technology topics from their unique generational perspective where the father often introduces cutting-edge tech to his millennial son rather than the reverse. The conversation spans from their experiences with Meta's Threads platform and its competition with X (formerly Twitter), to the evolution of AI from 1980s symbolic AI through today's large language models, and Microsoft's strategic shifts from serving programmers to becoming an enterprise-focused company. They also explore the historical development of search technologies, ontologies, and how competing technologies can blind us to emerging possibilities, drawing connections between past computing paradigms and today's AI revolution. To learn about Stewart Alsop II’s firsthand experience with Threads, check out his Substack at salsop.substack.com.Timestamps00:00 Stewart III shares how his dad unusually introduces him to new tech like Threads, reversing typical millennial-parent dynamics05:00 Discussion of Stewart's Chinese hardware purchase and Argentina's economic challenges with expensive imports and subsidies10:00 Analyzing Twitter's transformation under Musk into a digital warlord platform versus Threads serving normal users15:00 Threads algorithm differences from Facebook and Instagram, photographer adoption, surpassing Twitter's daily active users20:00 Threads provides original Facebook experience without ads while competing directly with Twitter for users25:00 Exploring how both Musk and Zuckerberg collect training data for AI through social platforms30:00 Meta's neural tracking wristband and Ray-Ban glasses creating invisible user interfaces for future interaction35:00 Reflecting on living in the technological future compared to 1980s symbolic AI research limitations40:00 Discussing symbolic AI, ontologies, and how Yahoo and Amazon used tree-branch organization systems45:00 Examining how Palantir uses ontologies and relational databases for labeling people, places, and things50:00 Neuro-symbolic integration as solution to AI hallucination problems using knowledge graphs and validation layers55:00 Google's strategic integration approach versus OpenAI's chat bot focus creating competitive pincer movementKey Insights1. Social Media Platform Evolution Through AI Strategy - The discussion reveals how Threads succeeded against Twitter/X by offering genuine engagement for ordinary users versus Twitter's "digital warlord" model that only amplifies large followings. Zuckerberg strategically created Threads as a clean alternative while abandoning Facebook to older users stuck in AI-generated loops, demonstrating how AI considerations now drive social platform design.2. Historical AI Development Follows Absorption Patterns - The conversation traces symbolic AI from 1980s ontology-based systems through Yahoo's tree-branch search structure to modern neuro-symbolic integration. Nothing invented in computing disappears; instead, older technologies get absorbed into new systems. This pattern explains why current AI challenges like hallucinations might be solved by reviving symbolic AI approaches for provenance tracking.3. Enterprise vs Consumer AI Strategies Create Competitive Advantages - Microsoft's transformation from a programmer-focused company under Gates to an enterprise company under Satya exemplifies strategic positioning. While OpenAI focuses on consumer subscriptions and faces declining signups, Anthropic's enterprise focus provides more stable revenue. The enterprise environment makes AI agents more viable because business requirements are more predictable than diverse consumer needs.4. Integration Beats Best-of-Breed in Technology Competition - Google's recent AI comeback demonstrates the Microsoft Office strategy: integrating all AI capabilities into one platform rather than forcing users to choose between separate tools. This integration approach historically defeats specialized competitors, as seen when Microsoft Office eliminated WordPerfect and Lotus by bundling everything together rather than competing on individual features.5. Technology Prediction Limitations and Pattern Recognition - The discussion highlights how humans consistently fail to predict technology developments beyond 2-3 years, while current developments within 12 months are predictable. This creates blind spots where dominant technologies (like transformers) capture all attention while other developments (like the metaverse) continue evolving unnoticed, requiring pattern recognition skills that current AI lacks due to reliance on historical data.6. Network Effects Transformed Computing Fundamentally - The shift from isolated computers with small datasets in the 1980s to today's high-speed global networks created possibilities unimaginable to early AI researchers. This network transformation explains why symbolic AI failed initially but might succeed now, and why companies like Palantir can use ontologies effectively with massive connected datasets that weren't available during the 1980s AI bubble.7. Professional Identity Boundaries Shape Technology Adoption - The distinction between hobbyist programmers seeking creative expression and IT professionals whose job is to "say no" and maintain standards reveals how professional roles influence technology adoption. This dynamic explains both historical patterns (like the Apple vs enterprise IT conflicts) and current challenges (like Microsoft Copilot adoption issues), showing how organizational structures affect technological progress beyond pure technical capabilities.

Ver todos los episodios