This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Catching bad content in the age of AI
Big Tech is surprisingly bad at catching, labeling, and removing harmful content. In theory, new advances in AI should improve our ability to do that. In practice, AI isn’t very good at interpreting nuance and context. And most automated content moderation systems were trained with English data, meaning they don’t function well with other languages.
The recent emergence of generative AI and large language models like ChatGPT means that content moderation is likely to become even harder.
Whether generative AI ends up being more harmful or helpful to the online information sphere largely hinges on one thing: AI-generated content detection and labeling. Read the full story.
Tate’s story is from The Technocrat, her weekly newsletter giving you the inside track on all things power in Silicon Valley. Sign up to receive it in your inbox every Friday.
If you’re interested in generative AI, why not check out:
+ How to spot AI-generated text. The internet is increasingly awash with text written by AI software. We need new tools to detect it. Read the full story.
+ The inside story of how ChatGPT was built from the people who made it. Read our exclusive conversations with the key players behind the AI cultural phenomenon.
+ Google is throwing generative AI at everything. But experts say that releasing these models into the wild before fixing their flaws could prove extremely risky for the company. Read the full story.
Podcast: Harvesting the future with AI and satellites
In the latest episode of our award-winning podcast, In Machines We Trust, the team travels from test farms to labs to learn how AI is used in farming in some ways you might not expect, like for tracking the health of crops—from space. Listen on Apple Podcasts or wherever you get your podcasts, and check out the first installment of this mini series if you haven’t already.
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Chatbots are more influential than you realize
They’re able to subtly alter our opinions in a similar way to other humans. (WSJ $)+ Today’s chatbots are just the tip of the AI iceberg. (Wired $)+ ChatGPT is going to change education, not destroy it. (MIT Technology Review)
2 Russia’s aircraft are full of US technology
Even though sanctions are supposed to prevent US goods from entering the country. (NYT $)
3 Gory news footage is only ever a few clicks away on social media
But it’s up for debate whether seeing it serves any real purpose. (WP $)
4 Sam Altman’s Worldcoin is nearing $100 million of funding
Despite some pretty major privacy and ethical issues. (FT $)
+ Learn about the dodgy methods Worldcoin used to recruit its first half a million test users. (MIT Technology Review)
5 Students are the US semiconductor industry’s secret weapon
Companies need highly skilled engineers, and universities are happy to oblige. (IEEE Spectrum)
7 What’s happening with Google Assistant, exactly?
The once-ubiquitous AI helper was noticeably absent from its I/O conference. (Wired $)
+ Google is prompting us how to prompt its own AI models. (The Verge)
+ That wasn’t Google I/O — it was Google AI. (MIT Technology Review)
8 Fire-suppressing chemicals are toxic to wildlife
A milestone lawsuit could stop emergency services from using it in the future. (The Atlantic $)
+ The quest to build wildfire-resistant homes. (MIT Technology Review)
9 Meet the internet’s favorite metal detectorists
The thrill of unearthing buried treasure is rife among online communities. (The Guardian)
10 Amazon is trying to build a smarter home robot
If it works, it’ll be a souped-up version of its Astro robot with ChatGPT-style smarts. (Insider $)
Quote of the day
“If you can survive here you can make it anywhere.”
—Tech entrepreneur Rosie Zhang tells the Financial Times how China’s intensely competitive environment helps its native companies to strike it big overseas.
The big story
If we want to be able to fix bias in AI, we need to understand the mechanics of how it arises in the first place.
We often shorthand our explanation of AI bias by blaming it on biased training data, but the reality is more nuanced. Bias can creep in long before the data is collected as well as at many other stages of the deep-learning process—and can be incredibly hard to fix. Read the full story.
We can still have nice things
+ If you’re enjoying The Legend of Zelda: Tears of the Kingdom, here’s a list of magical movies that compliment its vibe perfectly.
+ Be still my beating heart: Dolly Parton is getting into rock.
+ California’s superbloom is really very beautiful.
+ These adorable baby goats aren’t just cute—they’re also accomplished firefighters (thanks Anna!)
+ A little reminder of the importance of the ‘third’ places we spend time in outside of work and home.