The Trump administration is now ordering hospitals to send coronavirus patient data to a database in Washington, DC as part of a new initiative that may bypass the Centers for Disease Control and Prevention (CDC), according to a report from The New York Times published on Tuesday. The Verge reports: As outlined in a document (PDF) posted to the website of the Department of Health and Human Services (HHS), hospitals are being ordered to send data directly to the administration, effective tomorrow, a move that has alarmed some within the CDC, according to The Times. The database that will collect and store the information is referred to in the document as HHS Protect, which was built in part by data mining and predictive analytics firm Palantir. The Silicon Valley company is known most for its controversial contract work with the US military and other clandestine government agencies as well as for being co-founded and initially funded by Trump ally Peter Thiel.

"A unique link will be sent to the hospital points of contact. This will direct the [point of care] to a hospital-specific secure form that can then be used to enter the necessary information. After completing the fields, click submit and confirm that the form has been successfully captured," reads the HHS instructions. "A confirmation email will be sent to you from the HHS Protect System. This method replaces the emailing of individual spreadsheets previously requested." While the White House's official reasoning is that this plan will help make data collection on the spread of COVID-19 more centralized and efficient, some current and former public health officials fear the bypassing of the CDC may be an effort to politicize the findings and cut experts out of the loop with regard to federal messaging and guidelines, The Times reports.

Verizon has turned to Google Cloud's Contact Center AI to automate its customer-service phone calls and chatbot conversations. The Register reports: The Contact Center AI technology will, we're told, use natural-language recognition to transcribe on-the-fly customers as they talk down the line. This speech-to-text output will be fed into Dialogflow, a platform that parses the text and generates responses in real-time. Similarly, written conversations with online support chatbots will be processed in real-time by Google's AI. The overall aim is to allow subscribers to rant, er, complain away using natural language at the computer system, from their keyboards or over the phone, and the artificial intelligence should be able to work out what customers want, and help them out, without them having to navigate a menu or bark simple commands.

And presumably the aim is to sort out simple stuff quickly without a human operator having to come on the line and deal with it. Subscribers with trickier problems should also, we hope, be directed to a human being without having to negotiate their way through a menu or a script of irrelevant procedures. The software agents can also suggest relevant online documentation, such as information on how to view or pay a bill, based on a subscriber's request. Amusingly, if you get through to a human, or demand to speak to a person, the staffer will probably just tell you what the AI wanted to say anyway: the software will, behind the scenes, provide prompts to the call-center workers.