Good morning! Today: our climate change editor explains why he's not leaving California, a new AI that gets its facts straight by reading the entire web nonstop, the problems with Facebook's political ad ban, and use Google's AI to create your own moody quarantine music. Get your friends to sign up here to get The Download every day. | In defense of California About a year after graduating from college, I packed my possessions into a rental van I’d split with a near stranger and departed my home state of Ohio. We steered onto I-70 West, bound for San Francisco, writes our climate change and energy editor James Temple. At the time, I was less drawn to California in any specific way than determined to escape a state that was too conservative, homogenous, and religious for my tastes. Plus, oof, the winters. But that soon changed. The more I explored California’s coastline, hiked the trails of the Sierra, stared up at the granite walls of Yosemite, and met others who felt pushed or pulled here, the more I developed what I jokingly call a “zeal of the convert” attitude toward the state. Today, more than two decades after I arrived here in that rental van, this allegiance manifests as knee-jerk defensiveness when others take shots at California. And so it’s been heartbreaking to watch my adopted state suffer through some of the deadliest and most devastating fire seasons in its history. And it’s been infuriating to see commentators pounce on the tragedies, or the planned electricity blackouts designed to prevent them, and declare that they’ll doom the state or spark a mass exodus. It’s an increasingly popular take, but it runs into two questions: where else would people go? And in this moment, which place feels a whole lot safer? Read his full essay.
| | This know-it-all AI learns by reading the entire web nonstop The issue: Back in July, OpenAI’s latest language model, GPT-3, dazzled with its ability to churn out paragraphs that look as if they could have been written by a human. The trouble is that GPT-3 can be a bit of a bullshitter. Language models like GPT-3 are amazing mimics, but they have little sense of what they’re actually saying. This is a problem if we want AIs to be trustworthy. A new approach: Diffbot is doing things differently. It is building an AI that reads every page on the entire public web, in multiple languages, and extracts as many facts from those pages as it can. Each of these factoids gets joined up with billions of others in a sprawling, interconnected network of facts. This is known as a knowledge graph, and it’s a technique that underpins much of Google Search. Diffbot crawls the web nonstop and rebuilds its knowledge graph every four to five days. Still, even an AI that has its facts straight is not necessarily smart—just useful. Read the full story. —Will Douglas Heaven | Create your own moody quarantine music with Google’s AI AI fun times: Lo-Fi Player is the latest project out of Google Magenta, which makes machine learning tools for the creative process. It lets you mix tunes with the help of AI by interacting with a virtual room. What’s it about? The goal is to make the music-mixing experience as simple and friendly as possible. The room is a two-dimensional, pixelated drawing displayed in a web browser. Clicking on different objects, like the clock and the piano, prompts the user to adjust different tracks, like the drum line and melody. Give it a go for yourself. —Karen Hao | | The top ten must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 Why Facebook’s political-ad ban is taking on the wrong problem Its content recommendation algorithms do far more damage than targeted ads. ( TR) + The ban will make it much harder for officials to give people accurate information on how to vote. ( ProPublica) + Just telling people to delete Facebook really doesn’t help. ( The Verge) + Trump is not a Facebook user, but he thinks like one. ( WP $) + Why Facebook is going to war in Australia. ( Slate) 2 Rapid at-home testing won’t be a covid-19 panacea It’s a cool idea, but there’s still no real-world data to show it would work. ( NYT $) + The questions scientists are asking about coronavirus reinfections. ( Nature) + The UK reported a significant spike in new infections yesterday. ( BBC) 3 Detroit police wrongfully arrested another Black man due to facial recognition And he’s suing. ( Vice) + “Geofence” warrants are being used by police to indiscriminately sweep up data on anyone who happened to be near a crime scene. ( Wired $) 4 Russia has published data on its coronavirus vaccine 💉 All participants produced an antibody response—but it’s too early to say if it’s safe. ( Reuters) 5 Amazon UK is riddled with fake reviews The most prolific reviewer assessed £15,000 worth of products last month. He received most of them for free. They all got five stars. ( Ars Technica) + Amazon has banned foreign sales of seeds in the US after people kept receiving mysterious unsolicited packages. ( The Guardian) 6 Tech companies can’t fix racism without white people Leaving Black people to shoulder the burden is unfair—and doesn’t tackle the real issue. ( TR) 7 The pandemic is no excuse to spy on students Excessive surveillance almost always backfires. ( The Atlantic) 8 Eight case studies on regulating biometric technology show us a path forward A new report reveals which regulatory approaches help to protect communities from surveillance. ( TR) 9 Not everything about 2020 is awful There are plenty of changes to our lives that we’d like to make permanent. ( Axios) + The four-day week, for example. ( Wired UK) 10 TikTok is leading a skateboarding revival 🛹 And it’s more inclusive and supportive than ever. ( The Verge) | | “Very unlikely.” —Moncef Slaoui, chief adviser for the US coronavirus vaccine development program, tells NPR about the chances a vaccine will be ready to distribute by November. | | | | | |
No comments:
Post a Comment