Plus, Meta has been accused by some of burying the research on VR’s dangers.
Here is the latest edition of The Download,which is our weekly newsletter that gives you a daily dose on what’s happening in the worlds of technology.
Video: AI, our energy future. In May, MIT Technology Review released an unprecedented and comprehensive analysis of how much energy is used by the AI industry–down to a specific query. Our reporters and editors mapped out the current carbon footprint of AI and its future trajectory as AI continues to grow and reach billions of users.
A short video has been produced to accompany this investigation. You can read the entire story here and watch the video on YouTube.
AI changes the grid. Could it do more good than harm?
The growing popularity of AI is driving a rise in electricity demand that could reshape our grid. Data centers have increased their energy consumption by 80% between 2020 and 2025, and this trend is likely to continue. Electricity prices have already risen, especially in areas where data centers are concentrated.
But many people, particularly in Big Tech argue that AI is, on the whole, a positive force to the grid. They claim that this technology can help bring more clean energy online faster, run the power system more efficiently and predict and prevent failures which cause blackouts. What is the merit of that argument?
–Casey Crownhart.
We still don’t understand the energy burden of AI.
– James O’Donnell.
When my colleague Casey Crownhart, and I, spent six months researching climate and energy burdens of AI earlier this year, we found one number that stood out as our “white whale”: how much energy leading AI models like ChatGPT and Gemini use to generate a single answer.
Google, OpenAI and Microsoft refused to give us their figures for our article. Then, this summer, after our article was published, strange things started to happen. They started releasing the numbers that we had been demanding. Is our job done with this newfound transparency? Did we finally catch our white whale? I contacted some of our older sources and some newer ones to find out. Read the complete story.
MIT Technology Review Narrated : Google DeepMind’s new way to see inside an AI’s mind.
Neither do we know how AI works or why it is so effective. This is a problem, as it could lead to us deploying an AI system into a highly sensitive area like medicine without fully understanding its flaws. But a team from Google DeepMind has been working to find new ways to allow us to peer under the hood. This is the latest story that will be made into a MIT Technology Review narrated podcast. We’ll publish it each week on Apple Podcasts or Spotify
. Follow us on either platform and navigate to MIT Technology Review to receive all of our new content.
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
Meta suppressed the research into the harms that young users experience in VR
Former employees told a Senate Committee the firm did this to avoid regulatory scrutiny. (WP$)
2 The MAGA movement has a lot of AI skeptics.
However, the White House is removing regulatory barriers to speed up AI adoption. (FT $)
3 Pfizer claims its new covid vaccination boosts immune response fourfold
if you can get it. (Ars Technica, )
+ Americans without access to a booster vaccine are growing more and more fearful. The current vaccine guidance is confusing. Why limited access to Covid vaccines may not be a bad thing. MIT Technology Review
– 4 The EU will consider banning social media sites for those under 16
– Following governments in Europe who have pushed for mandatory age restrictions. (Bloomberg$)
4 RFK Jr. is all-in with ChatGPT
: All US health department staff have access to the tool. (404 Media)
+ Humans are more likely to accept disinformation that is generated by AI. MIT Technology Review
Six An “AI-supported coder” won a hackathon against a machine
But AI Tools seem to slow some experienced human developers. (Wired$)
+ We are in the second wave of AI-based coding. MIT Technology Review
Mark Zuckerberg is suing Meta.
No this Mark Zuckerberg. (NOW $)
+ A bankruptcy lawyer is tired of being mistaken as him. The Guardian
Apple’s AirPods can convert languages in real-time
via a robotic voice. (Ars Technica, )
+ New AI translation system for headphones clones several voices simultaneously. MIT Technology Review
AI is threatening Latin America’s diverse music scene
The proliferation of fake songs on streaming platforms is robbing artists of income. (Rest Of World )
+ Pandora’s streaming failure. Fast Company($)
+ How Spotify’s algorithm can be beaten. MIT Technology Review
Christie’s is axing their digital art division.
But, don’t worry – it’ll still sell NFTs. (Cointelegraph)
+ I tried to buy an Olive Garden NFT. All I got was heartburn. MIT Technology Review
Today’s Quote
“If Meta’s AI division doesn’t get the right foundation culturally, you will burn out a lot of people and piss them off. A lot of them will quit, and you will waste millions of money.” –Laszlo Bock a tech industry advisor and former head of Google’s people operations, explains to the Wall Street Journal where Meta’s AI is going wrong[19459032[19459032
Another thing
The $100 billion bet on a postindustrial US town reinventing itself as a hub of high-tech
A small drilling rig is positioned at the edge of the scrubby, overgrown fields in Syracuse, New York to take soil samples. It’s the beginning of construction for what could be the largest semiconductor manufacturing plant in the United States.
The CHIPS and Science Act, which was passed in 1965, was widely viewed as a way for industry leaders and politicians to secure supply chain and make the United States more competitive in semiconductor chip manufacture.
Syracuse is now about to be an economic test that will determine whether aggressive government policies and the massive corporate investment they spur can both boost the nation’s manufacturing prowess over the next few decades. Read the complete story.
