Daily Dish 072718

How U.S. tech powers China’s surveillance state [Axios]

  • Thermo Fisher Scientific, a Massachusetts company, has supplied the Chinese government with DNA sequencers that it is now using to collect the DNA of ethnic minorities in Xinjiang, Human Rights Watch reports. At a Thursday hearing, Sen. Marco Rubio called Thermo Fisher’s operations in Xinjiang “sick.”
  • iFlyTek is a Chinese company that recently launched a 5-year partnership with the Massachusetts Institute of Technology. Beijing has used iFlytek’s voice recognition technology “to develop a pilot surveillance system that can automatically identify targeted voices in phone conversations,” according to Human Rights Watch.
  • Cisco, in 2011, developed a network of 500,000 cameras that China used for street surveillance.

Thoughts On Machine Learning Accuracy [Amazon] 

Amazon responds to false positives .…The ACLU has not published its data set, methodology, or results in detail, so we can only go on what they’ve publicly said. But, here are some thoughts on their claims: ..

Strategic Competition in an Era of Artificial Intelligence [cnas.org]

The sharper the competition, though, the greater the need to also think about the potential for a race to the bottom in AI safety.

Vivienne Ming: ‘The professional class is about to be blindsided by AI’ [FT]

“If I take a deep neural network and throw it at a bunch of hiring data for the last 10 years, it will tell me to hire straight white men from wealthy conservative backgrounds,” she says. “There are many published economics papers about this, peer-reviewed. But in any big data set, there are two types of relationships that the AI can learn from. One is the spurious correlations, like the historical fact that white men hold most jobs. And then you have actual causal relationships such as the fact that successful people are resilient and have a growth mindset, social skills, emotional intelligence, creativity, meta-cognition and so on.

Transcript: Kara Swisher’s chat with Jaron Lanier

Yeah, I very strongly feel that we can isolate the good parts of social media which are very real and very true and just cut off and incinerate the bad parts, and the bad parts can be described very clearly as a manipulation engine. It’s the algorithms that are measuring you and then calculating what you should experience in order to change your behavior according to an algorithm. It’s that manipulation engine that’s the problem. It’s not the smartphone. It’s not the general idea of social media. It’s not the general idea of the internet. It’s none of those things. It’s really the manipulation machine. And that’s the thing that needs to be shut down.

How TV Tuned in More Ad Dollars: Drug Money, Digital Doldrums Kept Madison Ave. Attention on Linear Viewers [Variety]

The nation’s five English-language broadcast networks prevailed against disruptive trends brought on by new streaming-video technology and managed to snare a gain of between 3% and 5% the volume of advance advertising commitments they secured for their next cycle of primetime programming, according to Variety estimates, part of the annual haggling of TV’s “upfront” ad-sales marketplace. The networks secured between $9.1 billion and $10.06 billion, according to Variety estimates, compared with $8.69 billion and $9.55 billion in last year’s haggle

Machine Learning in Google BigQuery [Google AI]

Today we’re announcing BigQuery ML, a capability inside BigQuery that allows data scientists and analysts to build and deploy machine learning models on massive structured or semi-structured datasets. BigQuery ML is a set of simple SQL language extensions which enables users to utilize popular ML capabilities, performing predictive analytics like forecasting sales and creating customer segmentations right at the source, where they already store their data. BigQuery ML additionally sets smart defaults automatically and takes care of data transformation, leading to a seamless and easy to use experience with great results.