In a big upgrade to the “chat” features of Workplace (conversations that happen outside the news feed, first launched last year), users will now be able to start chats, calls and video conversations either one-to-one or in groups, in the style of WhatsApp or Messenger. Facebook is also making it easier to navigate through high volumes of messages in your channels by adding in replies, do not disturb and pinning features — Facebook’s first move to bring in algorithmic sorting to Workplace. And Facebook is also bringing its Safety Check feature from the main app to Workplace, delivered via Workchat, as a tool that can be controlled by admins to check on the status of employees during a critical incident.
Workplace has had multi-organizational chat for some time, but they are now offering voice or video calling without having to use a separate app. So, headed in a real-time communications direction, and in a way that supports communications across different companies.
in his coverage of the event
makes the smart observation that many of the 30,000 companies now using Workplace have highly mobile workforces, like Delta
One thing a lot of these companies have in common, other than being large, is that they have lots of employees who don’t sit in front of a PC. Instead, they’ve got workers who are up and about and access Workplace on a phone. Julien Codorniou, the Facebook VP in charge of Workplace, told me that this use-case scenario turned out to be more powerful than Facebook initially realized. And the the fact that Workplace has found traction outside the bubble—rather than as, say, a Slack archrival—might help explain why it hasn’t attracted that much attention.
It makes sense that Facebook would be pulled into the direction of workforce communications – a mobile-first UX – rather than the Slack work chat UX – a PC-first UX.
First, the legal set up that he attributes to Shawn Bayern:
Giving AIs rights similar to humans involves a technical lawyerly maneuver. It starts with one person setting up two limited liability companies
and turning over control of each company to a separate autonomous or artificially intelligent system. Then the person would add each company as a member of the other LLC. In the last step, the person would withdraw from both LLCs, leaving each LLC – a corporate entity with legal personhood – governed only by the other’s AI system.
That process doesn’t require the computer system to have any particular level of intelligence or capability.
Yampolsky runs down a long list of potentially awful outcomes – like malevolent AI trying to exterminate us, or gaining voting rights – but I’m more interested in the self-managing aspect of AI personhood. For example, the notion of responsibility of autonomous cars can be partially offset by using Bayern’s approach to AI personhood, creating a pair of LLCs that owns and operates the autonomous car, charges for its use, and – most importantly – pays for its own insurance and maintenance.
The great majority of the negatives Yampolsky enumerates could be handled by passing laws that would not allow AI to assume elected office, or enslave people.