The Atlantic


Astro Teller on Why Artificial Intelligence Is Not Scary

The Atlantic

"Taking over the world is an intensely human thing to want to do," says Astro Teller, in a short interview conducted at the 2017 Aspen Ideas Festival. At Google X, Teller studies and develops artificial intelligence. Here, he argues that current frenzy over the topic might be overblown.


What Should We Call Silicon Valley's Unique Politics?

The Atlantic

The "do not regulate" category was formed from responses to questions about regulating Uber, how the gig economy should be structured, whether it is too hard to fire workers, and the general proposition of whether "government regulation of business does more harm than good," as well as specific questions about regulating drones, self-driving cars, and internet companies. For example, 80 percent of tech founders think economic inequality is fine if it means the economy grows faster and 75 percent of tech founders think labor unions should lose influence. And yet, when the researchers asked the tech founders about taxation and redistribution policies, they expressed major support for things like "universal healthcare, even if it means raising taxes," increases in spending on the poor, and taxes on high-income individuals. If tech founders had their way, government regulation might not stop you from financially falling through market action, but it'd bounce you back up.


Inside Waymo's Secret World for Training Self-Driving Cars

The Atlantic

In a corner of Alphabet's campus, there is a team working on piece of software that may be the key to self-driving cars. No journalist has ever seen it in action until now. Months ago, a self-driving car team encountered a roundabout like this in Texas. Here, a single real-world driving maneuver--like one car cutting off the other on a roundabout--can be amplified into thousands of simulated scenarios that probe the edges of the car's capabilities.


The Moral History of Air-Conditioning

The Atlantic

In 1851, a Florida doctor named John Gorrie received a patent for the first ice machine. He'd been trying to alleviate high fevers in malaria patients with cooled air. To this end, he designed an engine that could pull in air, compress it, then run it through pipes, allowing the air to cool as it expanded. It wasn't until the pipes on Gorrie's machine unexpectedly froze and began to develop ice that he found a new opportunity.


How Uber Is Building Uber for Trucking

The Atlantic

As Uber battles taxis and other ride-hailing apps in cities across the world, the company is beginning to move quickly into a much larger transportation market: trucking. This spring, Uber unveiled Uber Freight, a brokerage service connecting shippers and truckers through a new app. Since then, the teams have split up into self-driving research and development, managed by Alden Woodrow, formerly of Google X, and the Uber Freight team. Even in trucking, Uber's acquisition of Otto has led to a lawsuit filed by Alphabet's self-driving car division, Waymo, related to the alleged theft of sensor technology.


Artificial Intelligence Is Just Another Talking Point

The Atlantic

Elon Musk and Mark Zuckerberg are having a spat about whether or not artificial intelligence is going to kill us all. "But until people see robots going down the street killing people, they don't know how to react." In a Facebook Live broadcast, Zuckerberg, Facebook's CEO, offered riposte. Seeing the CEOs of publicly traded tech companies go at it like Tay and Kanye is unfamiliar territory.


The Technology That Will Make It Impossible for You to Believe What You See

The Atlantic

Obama was a natural subject for this kind of experiment because there are so many readily available, high-quality video clips of him speaking. In order to make a photo-realistic mouth texture, researchers had to input many, many examples of Obama speaking--layering that data atop a more basic mouth shape. The researchers used what's called a recurrent neural network to synthesize the mouth shape from the audio. Recurrent neural networks are also used for facial recognition and speech recognition.)


All the Promises Automakers Have Made About the Future of Cars

The Atlantic

As part of that, they've promised to "bring highly automated driving functions to market as a core competency from 2021." They announced they're rolling out "Level 3" automation--which means a car that can drive itself some of the time--in the A8 model this year with promises to bring fully autonomous vehicles to market in 2020. On the electric side, the company has promised a sporty little electric vehicle called the I.D. Instead, the company's engineers had built them to run artificially well under testing conditions (and only under testing conditions).


Why Would Anyone Fear a Self-Driving Car?

The Atlantic

"Planes fly roughly 99 percent of the miles that they fly by computer. It's now to the place that it is not safe for humans to fly in a lot of conditions. If you could have a robotic surgeon that makes one mistake in 10,000, or a human that made one mistake in 1,000, are you really going to go under the knife with the human? As a counterpoint, however, there are lots of Americans who choose to drive rather than fly, fearing the latter more despite knowing that it is statistically much safer.


An Artificial Intelligence Developed Its Own Non-Human Language

The Atlantic

In the report, researchers at the Facebook Artificial Intelligence Research lab describe using machine learning to train their "dialog agents" to negotiate. At one point, the researchers write, they had to tweak one of their models because otherwise the bot-to-bot conversation "led to divergence from human language as the agents developed their own language for negotiating." In other words, the model that allowed two bots to have a conversation--and use machine learning to constantly iterate strategies for that conversation along the way--led to those bots communicating in their own non-human language. Already, there's a good deal of guesswork involved in machine learning research, which often involves feeding a neural net a huge pile of data then examining the output to try to understand how the machine thinks.