- Mindstream
- Posts
- Google's big emissions problem
Presented by
Happy 4th of July to our American friends - and happy Thursday to everyone else.
Despite its vast ice fields, Antarctica is a desert due to its low annual precipitation - just 2 inches.
What’s in store:
Google: From green dreams to carbon nightmares.
Figma pauses their new AI tool after showing similarities to Apple's App.
Around the web: Meta’s Threads hits 175M users in a year.
Live event from Vanta!
Read Time: 5 minutes
Google's latest environmental report reveals a significant rise in greenhouse gas emissions, complicating its climate goals amid its AI expansion.
Google has set a goal to halve its emissions by 2030 compared to 2019.
However, its greenhouse gas emissions have surged by 48% since 2019, hitting 14.3 million metric tons of CO2 last year alone – a 13% increase from the previous year.
This amount is comparable to the emissions from 38 gas-fired power plants annually.
The rise in emissions is mainly due to energy-hungry data centres and supply chain emissions.
Data centres, especially those used for AI, consume a vast amount of energy.
In 2023, electricity consumption from data centres added nearly a million metric tons of pollution to Google's carbon footprint.
Google's move towards integrating AI, such as redesigning Search with generative AI, has further driven up energy demands.
The environmental report acknowledges that incorporating AI increases emissions due to the higher energy needs of AI computing.
What you should know:
Google’s greenhouse gas emissions have increased by 48% since 2019.
In 2023, the company produced 14.3 million metric tons of CO2.
Data centre electricity consumption grew by 17% in 2023.
The cost of innovation
Google's data centres alone accounted for up to 10% of global data centre electricity consumption last year.
The company plans to improve the energy efficiency of its AI models, hardware, and data centres.
Additionally, Google aims to operate on carbon pollution-free energy 24/7 by 2030.
Globally, data centres consume about 1% of the world's electricity.
With the AI industry booming, the International Energy Agency (IEA) estimates this could increase tenfold by 2026.
In the US, the country with the most data centres, there's concern that rising electricity demands from AI could strain power grids and extend the use of coal and gas plants.
Google is not alone in this struggle. Microsoft's greenhouse gas emissions were 30% higher in its 2023 fiscal year than in 2020.
Despite the challenges, Google remains committed to its environmental goals, balancing its AI growth with sustainability efforts.
Who knew saving the planet would be so…energy-intensive?
Is Google’s AI expansion worth the environmental cost? |
Vote for live results and see results + opinions from yesterday at the bottom of the email.
LIVE EVENT
Join the live session: automate compliance & streamline security reviews
Whether you’re starting or scaling your company’s security program, demonstrating top-notch security practices and establishing trust is more important than ever.
Vanta automates compliance for SOC 2, ISO 27001, and more, saving you time and money — while helping you build customer trust.
And, you can streamline security reviews by automating questionnaires and demonstrating your security posture with a customer-facing Trust Center, all powered by Vanta AI.
On This Day
July 4th 1960 - The 50-star flag of the United States debuted in Philadelphia, a little over a year after Hawaii became the 50th state.
GPT of the Week
The AI Cover Letter Generator
The AI Cover Letter Generator assists users in creating customised cover letters by collecting their resumes and job links. It analyses the provided information to craft a personalised cover letter that aligns with the job requirements. Additionally, it offers to tailor users' resumes to enhance their job applications further.
FIGMA
Figma’s new AI tool, Make Designs, was pulled after generating app mockups that closely resembled Apple's iOS weather app.
Figma's CEO, Dylan Field, took responsibility, citing the pressure to meet deadlines.
Andy Allen from Not Boring Software pointed out how Figma’s tool produced near-replicas of Apple’s weather app.
He advised designers to check their results to avoid legal issues.
In an interview, Figma CTO Kris Rasmussen said their AI models weren't trained on Apple’s designs.
Instead, the issue stemmed from the third-party models they used. Rasmussen confirmed that no training was done on Figma’s part.
Field stated that the Make Designs feature wasn’t trained on Figma content or app designs.
He noted that the issue was a lack of design variability.
The key AI models behind Make Designs are OpenAI’s GPT-4 and Amazon’s Titan Image Generator G1.
This raises questions about whether these third-party models were trained on Apple’s designs.
Here are the latest updates:
Figma's AI tool created app designs similar to Apple's weather app.
Figma's AI models weren't trained on specific app designs, but used third-party models.
Figma aims to improve design variability and be transparent with AI training policies.
Innovation meets legal issues
Rasmussen explained that Figma avoided training its features until it could be open about its AI policies.
Users have until August 15th to opt in or out of allowing their content for AI training.
Rasmussen hinted at future plans to train their own models, focusing on general design patterns to aid professional designers.
To prevent similar issues, Figma will ensure greater design variation and maintain high standards.
He acknowledged that Make Designs is in beta and not perfect, but assured users that the tool would be re-enabled soon after further checks.
Other Figma AI features remain available in beta, accessible via a waitlist.
Figma isn't alone in facing scrutiny over AI in creative tools. Adobe and Meta have also had to address concerns about how they use content to train their AI systems.
That’s what we call a legal headache.
ICYMI
A year after its launch, Meta's Threads has reached 175 million monthly users, becoming a notable rival to Elon Musk’s X. Despite its steady growth, daily user numbers remain undisclosed, suggesting many users aren't regulars. Promoted within Instagram, Threads shows potential as Meta's next billion-user app, with future plans to introduce ads.
AI Art
Our Image of the Day
Artwork submitted by Mindstream reader Elliott B: “A group of cavemen discovering a futuristic tablet in a cave”
Daily Image Prompt
4th of July parade with floats and fireworks
Yesterday’s Poll
“Do you think Anthropic’s new benchmarking program will truly elevate AI safety standards?”
Yes, it's a step in the right direction. - 71% ✅
No, it's just another corporate gimmick. - 29%
Your Views:
“Safety standards and the benchmarks called out go beyond the AI itself. Leadership and organizational cultural components will be key to the successful implementation and on going monitoring/benchmarking of AI going forward.” - brooketompkins
“I hope they'll also use an independent organization to validate their findings to build the trust with the consumer that it's not just a marketing ploy.” - heather.brizzi
Submit your opinions in our daily poll to be featured!
 | Happy 4th! Make sure to come back for more of the best. ❤️ We need your feedback to make our newsletter better. 📣 Refer our newsletter to your friends and make us feel warm and fuzzy. 🚀 Advertise in our newsletter to reach 150,000+ founders, engineers, and content creators. |
You There! What Did You Think? |