- AI in 5
- Posts
- Apple Joins OpenAI Board, and Perplexity Enhances Search to Rival Google
Apple Joins OpenAI Board, and Perplexity Enhances Search to Rival Google
Albert Chan
July 04, 2024
Cloudflare Unveils Tool to Thwart AI Data Scraping Bots
Cloud service provider Cloudflare has introduced a new tool designed to prevent bots from scraping data on its hosted websites for training AI models or other malicious purposes. The tool employs machine learning and specialized bot detection models to identify AI bots attempting to mimic legitimate website users. This initiative aims to protect content owners from unauthorized AI scraping, especially by bots that bypass the robots.txt file. For example, over 600 news sites have blocked OpenAI’s bot for ignoring these rules to train its models.
Why it matters: This tool addresses growing concerns about AI bots bypassing restrictions and scraping data without permission, highlighting the ongoing battle between content protection and AI development.
Perplexity AI Enhances Search to Rival Google
AI search engine startup Perplexity has unveiled significant upgrades to its Pro Search feature, aiming to compete with Google. The enhanced engine can now link multiple searches to deliver comprehensive, multi-step information summaries for complex queries. For example, it provides detailed guidance on viewing the Northern Lights, including optimal times, locations, and top viewing spots. Additionally, the upgrade boosts its math and programming capabilities for better data analysis and code debugging, though it's currently US-only.
Why it matters: These advancements aim to challenge Google's dominance in search, despite recent accusations against Perplexity for unethical data scraping practices.
Google's AI Push Drives Up Emissions, Complicates Climate Goals
In its latest Environmental Report, Google revealed a 13% increase in greenhouse gas emissions last year, largely due to the energy demands of its AI data centers. Despite a 2021 pledge to achieve “net zero” emissions by 2030, emissions are now 48% higher than in 2019, equating to 14.3 million metric tons of CO2—comparable to the output of 38 gas-fired power plants. This trend mirrors the broader tech industry, with Microsoft's emissions also rising by about 30% last year.
Why it matters: As AI data centers consume more power, reversing climate change becomes increasingly challenging, highlighting the environmental impact of AI development.
Apple Gains Observer Role on OpenAI Board
As part of their new partnership announced at Apple's WorldWide Developer Conference, Apple has secured an observer role on OpenAI's nonprofit board, with App Store Chief Phil Schiller representing the company. Although Schiller lacks direct involvement in AI and will have no voting rights, his longstanding contributions to Apple earned him the ‘Apple Fellow’ title, making him a fitting representative. This role will commence later this year, enabling Apple to gain insights into OpenAI’s operations as they integrate ChatGPT into iOS and macOS.
Why it matters: With Microsoft already holding a similar non-voting position on OpenAI's board, this arrangement could heighten competition and complicate discussions on AI initiatives between the tech giants.
YouTube’s Removal of AI-Generated Content That Simulates Your Face or Voice
YouTube has introduced a new policy allowing individuals to request the removal of AI-generated content that mimics their face or voice. This update is part of YouTube's privacy request process and builds on its responsible AI agenda announced last November. The policy requires first-party claims, except in cases involving minors, deceased individuals, or others who cannot access a computer. The request does not guarantee removal, as YouTube will evaluate the complaint based on factors such as whether the content is labeled as synthetic, its public interest value, and if it involves a public figure in sensitive contexts.
Why it matters: This policy sets a precedent for how tech companies might manage AI's impact on personal identity and consent as the lines between reality and digital imagination continue to blur.
New Ternary Approach Could Slash AI Training Energy Use
Researchers at the University of California, Santa Cruz have developed a method that converts numbers in AI models into ternary values, significantly reducing the energy required for training. This technique replaces the energy-intensive matrix multiplication process with simpler calculations using 0, 1, and -1. They successfully ran a billion-parameter model using just 13 watts of power, equivalent to an LED light bulb, without sacrificing performance. While it's not yet compatible with trillion-parameter models, the researchers are optimistic about scaling up and adapting the method to existing hardware.
Why it matters: This breakthrough could drastically cut the environmental impact of AI, making training large models much more energy-efficient.
Runway's Gen-3 Alpha Text-to-Video Model Now Open to All
Runway AI has launched its Gen-3 Alpha video model, now generally available, showcasing significant advancements in fidelity, consistency, and motion over its predecessor. Trained on a large-scale multimodal infrastructure, Gen-3 Alpha allows users to create high-quality videos from text prompts, with examples ranging from sci-fi visuals to dynamic time-lapses. Social media users have demonstrated its capabilities, generating diverse and creative content like hyper-realistic visuals and controlled camera speeds.
Why it matters: This release underscores the expanding role of AI in enhancing visual communication and creative expression, pushing the boundaries of what's possible with text-to-video technology.
ElevenLabs Brings Iconic Voices of Hollywood Legends to Reader App
ElevenLabs has partnered with the estates of Judy Garland, James Dean, Burt Reynolds, and Sir Laurence Olivier to bring their iconic voices to its Reader App, enabling users to hear digital texts narrated by these stars. Exclusive to the app, these voiceovers provide an emotionally rich and context-aware experience. Launched last week, the app uses AI technology to transform digital text into engaging narrations, honoring the legacies of these actors and offering fans new ways to connect with their voices.
Why it matters: This innovative collaboration enhances the accessibility and enjoyment of digital content, allowing users to experience the timeless voices of legendary stars in modern formats.
Microsoft AI Boss Sparks Controversy Over Use of Online Content
Microsoft AI head Mustafa Suleyman stirred debate by stating that content published on the open web is "freeware" and can be freely copied and used. In an interview with CNBC, Suleyman suggested that the longstanding social contract treats open web content as fair use, a view conflicting with copyright laws that protect a creator's work automatically. His comments come amid lawsuits against Microsoft and OpenAI for allegedly using copyrighted online stories to train AI models, highlighting ethical and legal concerns in AI practices. Suleyman also mentioned the robots.txt file, a guideline rather than a legal document, which some AI companies reportedly ignore.
Why it matters: Suleyman's remarks underscore the ongoing tension and legal uncertainties in AI content usage, emphasizing the need for clearer ethical and legal frameworks.
Meta Updates Policy on AI-Generated and Manipulated Media
Meta has revised its handling of AI-generated and manipulated media on Facebook, Instagram, and Threads, adding "AI info" labels to a broader range of content when AI indicators are detected or disclosed. The new approach emphasizes labeling with context rather than removal, except for content that violates other standards or poses a high risk of public deception.
Why it matters: Meta's updated policy aims to balance transparency and free expression, addressing the evolving challenges of AI-generated content in social media.
