Home Technology Open-Source Tools The Download: AI in the classroom and combating audio deepfakes

The Download: AI in the classroom and combating audio deepfakes

0
The Download: AI in the classroom and combating audio deepfakes

Plus, Nvidia announces it will resume sales of AI chip to China

This is the daily edition of The Downloadour weekday newsletter, which provides a daily dose on what’s happening in the worlds of technology.

AI programs that convert text to speech could one day “unlearn”how to imitate specific people.

News: Using a new technique called “machine-unlearning”AI models could be taught to forget certain voices.

The process: Companies currently deal with this problem by checking if the AI’s responses or the prompts contain prohibited material. Machine unlearning asks instead if an AI can be made forget a piece information that the company does not want it to know. It works by using a model, and the data that needs to be redacted to create a brand new model. This is a version of the initial that has never learned the data.

What’s the point? It could be a major step in preventing audio deepfakes where someone’s voice has been copied to commit fraud or scams. Read the complete story.

–Peter Hall.

AI giants want to take control of the classroom.

It’s summer and school is out, but a group of teachers are planning how they will use AI in the upcoming school year. God help them.

OpenAI, Microsoft and Anthropic announced on July 8 a $23-million partnership with one of America’s largest teachers’ unions to bring AI into K-12 classrooms. They will begin training teachers in New York City this fall on how to use AI for teaching, as well as for tasks such as planning lessons and writing report. These companies may face an uphill struggle. There is no clear evidence that AI will benefit students. It’s also hard to believe that the AI companies that are funding this initiative can give honest advice about when to and not use AI in the classroom. Read the complete story.

– James O’Donnell

Originally published in The Algorithm – our weekly newsletter on AI. Sign up here orto receive stories like these in your inboxfirst.

The must-reads

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

Nvidia claims that the US has lifted their ban on AI chip sale to China
Jensen Huang sweet-talked Donald Trump to reverse his three-month-old ban. (BBC)
+ This company will begin selling its H20 chip in China. WSJ ($)
+ America could impose tariffs on a material used in solar panels and chips. FT ($)

China has launched its digital identity system
This will give the country greater powers to monitor and censor their internet users. (WP $)

3 xAI has secured an agreement with the US Department of Defense just days after its Grok bot had a meltdown anti-Semitic. The Guardian]
+ EU officials hold talks with X representatives following the outburst. ($Bloomberg )

Four Meta’s data centres are on the brink of triggering a severe water shortage
Newton County, Georgia residents are suffering. (NYT $)
+ But, Zuckerberg still wants to build centers that are gigawatts in size. [Bloomberg[Bloomberg ]
+ Did you know that AI has a huge energy footprint? Here’s a story you may not have heard. MIT Technology Review

Rather then sending it to those in need, the Trump administration is burning tons of emergency food. The Atlantic $]

6 The US is trying to revive its rare earth industry
The Pentagon has invested over $1 billion in American company MP Materials. (WSJ $)
+ This is all part of a strategy to counter China’s mineral dominance. This rare earth metal reveals the future of our planet’s resources. MIT Technology Review

Seven AI nudifying applications are big business
These apps make millions of dollars a yea, and rely heavily on technology built by US companies. (Wired $)
+ Lensa, the viral AI avatar app, undressed without my consent. MIT Technology Review

Can anything save the internet at this point?”
Traffic has dropped, and AI usage is on the rise. (Economist $)
+ Fixing the internet. MIT Technology Review

Bytedance is developing its own mixed-reality goggles.
This comes a couple of years after the company scaled back work on an AR/VR headset. (The Information $)
+ The next step for smart glasses. MIT Technology Review

10 Minecraft’s spawned a new generation of entrepreneurs
This game encourages its players to learn programming. (Insider $)

Today’s Quote

I felt pure, unconditional Love.

–Faeight describes her strong feelings towards a previous AI partner.

Another thing

Making end-of-life decisions can be difficult and distressing. Could AI help with this? Surrogates, the people who make these decisions on behalf of others, can be very upset by end-of-life choices. It’s possible that family or friends disagree on what is best for a loved one. This can lead to stressful situations. David Wendler is a bioethicist working at the US National Institutes of Health. He and his colleagues are developing a tool based on artificial intelligence that can help surrogates to predict what patients would want in varying situations.

Wendler plans to build his tool as soon they receive funding, which could be in the next few months. It won’t be easy to roll it out. Critics question how AI can be ethically trained using a person’s personal data and if life-or-death decision-making should ever be left to AI. Read the complete story.

–Jessica Hamzelou

www.aiobserver.co

Exit mobile version