Google Verdict: Examining the Exhibits

The Google verdict last week generated significant buzz in the tech world. You can access all 400+ exhibits from the case, including internal email threads and testimonies, on Google Drive.

The Verge interviewed the Department of Justice (DoJ) lawyer responsible for the verdict, offering insights into the case.

Youtube Targeting Under-18s

The Financial Times reports that Instagram ran a user-acquisition campaign on Youtube targeting under-18s, which is against regulations. They achieved this by targeting users with "unknown" age, a segment that skews younger, in collaboration with Youtube.

This raises concerns about data privacy and platform responsibility in protecting minors.

AI & Spell Checking: A New Era of Software?

The author explores the evolution of spell checking from separate software to integrated features within operating systems. They draw parallels between the development of spell check and AI, questioning whether AI is simply a new stage in the evolution of software.

The author suggests that LLMs could be integrated into spell checkers and grammar checkers, creating "smart" or "auto" features that enhance writing.

However, they argue that LLMs might not be a universal solution for all tasks and that dedicated vertical products might emerge instead.

OpenAI's Moves: Investments and Resignations

OpenAI continues to be a major player in the AI world, making investments in startups like Opal, a webcam company. However, the company has also seen a number of resignations, with founders leaving to join Anthropic, a competitor founded by former OpenAI employees.

The author analyzes OpenAI's investment in Opal, questioning its long-term viability and the potential for AI to be integrated into hardware.

The AI Maximalist View: Will LLMs Be the Universal Solution?

The author discusses the "AI Maximalist" view, which posits that LLMs will eventually evolve to perform a wide range of tasks, including summarization, analysis, and even creating spreadsheets. They suggest that this is unlikely and that dedicated products and APIs might be more effective for specific use cases.

The author uses image recognition as an example of how AI technology needs to be integrated into structured workflows and user interfaces to become a viable product.

The Evolution of Tech: Features, Vertical Products, and General Purpose Experiences

The author concludes by outlining three models for deploying AI technology: features within existing products, dedicated vertical products, and a general purpose experience. They suggest that even a general purpose AI model might ultimately be a vertical experience due to the limitations of a blank canvas.

The author believes that the future of AI will likely involve a combination of these models, with specific products and APIs complementing the broader capabilities of LLMs.