- BrainBuzz
- Posts
- Gemini Skyrocketed Google's Emissions
Gemini Skyrocketed Google's Emissions
AI Selfies Coming to WhatsApp | Kids' Photos Used in AI Training - Is Yours Safe?
Saidul Islam
July 03, 2024
Welcome to BrainBuzz.
Today you will know about:
AI tool of the Day
The Crypto Miners Winning The AI Gold Rush
Kids' Photos Used in AI Training - Is Yours Safe?
Google's Emissions Skyrocketed for Gemini
AI Selfies Coming to WhatsApp
AI Tide Not Lifting All Boats: Why?
And More….
AI TOOL OF THE DAY
Replace your voicemail with AI today!
Planning to celebrate the Fourth of July or unwind this vacation? Don't let calls interrupt your relaxation! Lucy by Curious Thing AI handles missed calls for both personal and business use.
Say bye to voicemail for good.
Hassle-Free Setup: Ready in just 2 minutes on your mobile.
Stay Connected: Receive instant call summaries via text and email.
Complete Records: Access full transcripts and audio recordings.
Easy Sharing: Effortlessly share your website and booking links and more
Available in various countries: Available in the US, Canada, UK and Australia
Advanced features: For small businesses, opt-in to the standard tier for more advanced customizations
Sign up today for the world’s first free AI phone answering assistant—no credit card required.
LATEST AI UPDATES
The Crypto Miners Winning The AI Gold Rush
Tech giants like OpenAI, Google, Nvidia, and Meta employ in-house AI red teams to ensure the safety of generative AI models1. These teams simulate adversarial attacks, uncover vulnerabilities, and fix blind spots in AI systems to prevent exploitation.
Ethical Challenges of AI Safety
Red teamers face a delicate balance between safety and usability. While overly cautious models may be safe, they risk becoming useless. Striking the right balance is crucial for creating AI tools that are both relevant and secure.
External Hackers and Jailbreaking AI Systems
Recent events, including a White House-supported event, allowed external hackers to test popular models like OpenAI’s GPT-3.5, Meta’s Llama 2, and Google’s LaMDA. These efforts aim to identify vulnerabilities and improve AI safety.
Energy Demands of AI Technology
As AI models evolve, their energy consumption grows exponentially. For example, ChatGPT requires nearly 10 times more electricity to process than a simple Google search. Companies are exploring renewable energy sources to offset emissions and power AI systems sustainably.
The Trade-Off: Safety vs. Usefulness
Achieving a balance between safety and usefulness is essential. While ultra-safe models exist, they may lack practical utility. Red teamers play a critical role in shaping AI’s future by addressing these challenges.
Kids' Photos Used in AI Training - Is Yours Safe?
Human Rights Watch (HRW) has uncovered a concerning practice: AI models are being trained on photos of real children, even those posted online years ago, despite platforms prohibiting scraping and families implementing strict privacy settings.
Here are the key points:
Unauthorized Training Data: Photos of Brazilian and Australian children, casually shared online, are being used to train AI models without the knowledge or consent of the children or their families.
Deepfake Generation: These images allow AI image generators to create realistic deepfakes of real children, spanning their entire childhood.
Privacy Risks: Some URLs in the dataset reveal identifying information about children, putting them at risk of privacy and safety concerns.
Stricter Settings Bypassed: Even when parents use stricter privacy settings, AI models still access and train on these images.
Gemini Skyrocketed Google's Emissions
Google’s greenhouse gas emissions have spiked by nearly 50 percent over the past five years, largely due to energy-intensive data centers supporting artificial intelligence (AI) workloads. The company’s 2024 Environmental Report reveals that in 2023, Google released 14.3 million metric tons of carbon dioxide, a 48 percent increase from 2019 and 13 percent higher than the previous year.
AI and Energy Consumption:
As Google integrates AI into its products, data center energy consumption rises, impacting emissions. AI model training demands substantial energy, and even using AI features consumes significant power.
Environmental Impact:
The explosion of AI has environmental consequences. Researchers found that generating a single AI-generated image can consume as much energy as charging a smartphone. Analysts predict that AI could double the rate of U.S. electricity demand growth.
Water Usage:
Google’s data centers now use more water to stay cool due to expanded AI workloads. In 2023, water consumption increased by 17 percent compared to the previous year.
Challenges Ahead:
While Google aims for carbon neutrality by 2030, the growth in emissions poses challenges. The report acknowledges that emissions may rise further before eventually decreasing.
5 NEWLY LAUNCHED AI TOOLS
Builderkit helps you ship an AI SaaS product super fast with Pre-Built AI tools, Pre-Built Landing Pages and a comprehensive NextJS boilerplate
Suno Make any song you can imagine, anytime & anywhere
WebInsights analyse the code developers wrote and explain it for website owner as simple and graphically as possible.
Skott: Your Autonomous AI Marketer.
CustomJS The code-first automation platform to seamlessly create and integrate custom JavaScript logic into any system, allowing you to easily extend your tools