Mapping the AI Reality

Introduction

On 12 January 2025, Google and Apple released a joint statement announcing a landmark collaboration.

• This partnership aims to power future Apple Intelligence features, promising a more personalized Siri in the coming years.

Since its debut in October 2011, Siri has struggled to become the transformative, mainstream voice-controlled AI it initially promised to be.

• For years, it was relegated to rudimentary tasks - sending messages, setting reminders, and making calls - failing to evolve into a sophisticated "answering engine".

In today’s fast-paced world, users demand quick answers backed by advanced reasoning.

• After reportedly failing to reach an ideal long-term agreement with OpenAI regarding ChatGPT, Apple has pivoted toward utilizing Google’s Gemini models as a foundational pillar for its ecosystem.

The AI Integration Landscape

2025 has become the year of deep AI integration.

• Microsoft is aggressively embedding Copilot across its entire Office suite, while AI developers race to prove their models are both cost-effective and accurate, striving to eliminate "hallucinations" (false claims).

While ChatGPT pioneered the public AI craze in late 2022, it is arguably losing some of its initial luster as it faces astronomical operating costs and a search for sustainable profit margins.

• Meanwhile, Google has asserted its dominance in specific niches.
• Following the 2023 launch of NotebookLM and the industry-leading Nano Banana image editing tools, Google revealed it has successfully designed TPUs (Tensor Processing Units) that are significantly cheaper for large-scale AI than Nvidia’s GPUs.
• This is a strategic advantage, as competitors like OpenAI and Meta continue to spend billions on Nvidia hardware.

This boom has also spotlighted TSMC (Taiwan Semiconductor Manufacturing Company), the primary manufacturer for both Nvidia chips and Google TPUs.

• Furthermore, the massive demand for AI reasoning and data storage has surged requirements for high-capacity SSDs and HDDs.
• While the "Chatbot" business model is still finding its financial footing, the infrastructure sector - data center construction, electricity, and water utilities - is currently seeing the most consistent profits.

The AI Bubble and "Closed-Loop" Risks

Despite the excitement, there is growing caution regarding how tech companies will monetize these billion-dollar investments.

• The prevailing subscription model creates a state where users never "own" information but instead become permanently dependent on a service.
• This is analogous to university libraries paying high annual fees for digital access instead of owning physical books, or the Spotify model where music is rented rather than owned.
• The sustainability of this model is under fire, as many companies currently offer free access despite a net-negative cash flow.

While AI can generate code quickly, integrating it into complex systems often requires programmers to spend more time "reverse-engineering" the AI's logic to fix bugs than they would have spent writing it from scratch.

A more existential threat is the closed-loop data cycle.

• AI models feed on human-generated content from the web. If users stop creating original content and rely solely on AI-generated answers, the "well" of new knowledge will run dry.
• This mirrors the "Index Fund" dilemma in investing: index funds work by tracking the active research and valuations made by human traders.
• If every trader moves to an index fund, the market loses its price-discovery mechanism.
• Similarly, if human creators stop posting, AI will eventually have nothing new to learn.

Big Data and Governance

While AI eases daily tasks, it does so at the cost of unprecedented data capture.

• Every search and inquiry is harvested to build intrusive advertising profiles.
• AI models are shifting from "answering" to "acting" - recommending and booking hotels or trips directly.
• This creates a delusion of choice. Users are led to believe they are getting the best deal, when in reality, companies may "cherry-pick" results based on sponsored content or profit margins.

Furthermore, Google’s influence has expanded beyond its Chrome browser monopoly.

• Beyond guiding the development of the Android ecosystem, Gemini AI has gained a foothold in hardware from Samsung, Xiaomi, Oppo, and even Apple.
• With direct access to Gmail, Drive, and Photos, Google possesses a data goldmine that makes privacy increasingly fragile.
• While many users find Windows Copilot intrusive, Google has integrated similar capabilities across its services with far less public friction.

Summary

Artificial Intelligence has undeniably provided significant benefits, but a reality check is necessary: AI is a tool, not a replacement for human experience.

• True knowledge and skill are best acquired through the human interactions that define us.
• Much like Meta’s "Metaverse" struggled to gain traction because it lacked integration with the physical realities of life - paying bills and providing for families - AI must remain an assistant rather than the focal point.
• Our lives may be assisted by algorithms, but they should not be governed by them.