Zuckerberg Targets 'Personal Superintelligence'; Meta Shifts to Closed AI

Mark Zuckerberg, CEO of Meta, recently outlined the company’s ambitious new AI strategy: to “bring personal superintelligence to everyone.” This vision moves beyond simply automating business operations, aiming instead to empower individual goals and dreams through advanced AI assistants.

Zuckerberg suggested that future computing will center around “personal devices like glasses,” which can offer a rich, multimodal context for deep AI experiences. In a notable shift from Meta’s previous open-source approach, he also hinted that the company might exercise caution regarding which advanced AI models it makes publicly available, citing safety concerns related to superintelligence. This commentary aligns with recent reports that Meta paused work on its large open-source “Behemoth” model to redirect resources towards closed-source development. This strategic pivot is particularly significant as other nations, like China, continue to advance open-source frontier models. The announcement provides the most detailed insight yet into Meta’s “personal” AI vision, which aligns well with the company’s smart glasses initiatives.

In related AI news, Fable, a platform dubbed the “Netflix of AI,” has launched its Showrunner platform in Alpha, backed by an undisclosed investment from Amazon. Showrunner enables users to generate personalized, playable animated TV episodes using text prompts. Publicly launching this week, the platform will offer two original shows where users can influence narratives and create episodes within established worlds. Fable envisions a future of animation that is “remixable, multiplayer, personalized, and interactive,” allowing users to even upload themselves as characters. Initially free, the platform plans to introduce a monthly fee for generation credits and eventually enable revenue sharing for creators whose content is remixed. Showrunner first gained attention in 2023 for an experimental release of personalized, albeit unauthorized, South Park episodes. Its launch comes at a sensitive time for AI in the entertainment industry, potentially pioneering a new style of two-way, personalized content that could compel traditional intellectual property holders to re-evaluate their strategies towards user-generated content.

Meanwhile, Google DeepMind introduced AlphaEarth Foundations, an AI model designed to function as a “virtual satellite.” This model integrates vast amounts of Earth observation data from public sources, including optical images, radar, and 3D laser mapping, to create detailed, on-demand maps of the planet’s changing landscapes and coastal waters. AlphaEarth reportedly outperforms similar AI systems in terms of accuracy, speed, and efficiency, enabling near real-time tracking of environmental events like deforestation or ecosystem changes. Google has tested the dataset with over 50 organizations and plans to provide yearly updates via Earth Engine, facilitating the monitoring of long-term environmental shifts. This AI innovation significantly bridges the gap in processing disparate satellite data, transforming scattered feeds into unified maps that reveal previously imperceptible patterns.

In other developments:

• Anthropic is reportedly seeking to raise $5 billion in a new funding round led by Iconiq Capital, which would value the company at $170 billion, nearly tripling its valuation from March.

• OpenAI announced “Stargate Norway,” its first European data center initiative, a joint partnership with Aker and Nscale.

• YouTube is rolling out new AI-powered content moderation tools designed to estimate a user’s age based on viewing history and other factors, aiming to enhance the protection of minors.

• Neo AI debuted NEO, an “Agentic Machine Learning Engineer” powered by 11 agents, claiming state-of-the-art performance on ML-Bench and Kaggle competition tests.

• Amazon is reportedly paying between $20 million and $25 million annually to license content from The New York Times for AI training and use within its AI platforms.

• A recent study by The Associated Press found that the primary use of AI is for information searching, with young adults also frequently utilizing the tool for brainstorming.

• Among trending AI tools: Ideogram’s Character model allows placing specific characters into scenes; ChatGPT introduced Study Mode for guided learning; Writer launched Action Agent, an enterprise autonomous AI; and NotebookLM provides new video overviews that generate narrated slides.