Thread by @AngelSDiaz_

I'll be sharing a few thoughts from NYC's automated decision systems task force's first public forum. #NYCAlgorithms

Livestream here: nyls.mediasite.com/mediasite/Cata…

Co-Chair Kelly Jin notes that people from Australia are tuning in to the live stream.

This shows the extent to which communities around the world are looking to NYC to be a leader in algorithmic transparency and accountability. There's A LOT of work to do to meet that goal...

Futurist Andrew Nicklin recommends that NYC should impose requirements onto vendors to share the load of algorithmic accountability/transparency. YES. the city has lots of negotiation leverage, and can require all kinds of ongoing audits.

Nicely phrased: we vet elected officials responsible for making decisions about public welfare; we should have the ability to vet machines doing the same thing.

Sarah Kaufman, Associate Director, New York University Rudin Center for Transportation lays out the extent to which public movements across the city are tracked, logged, and shared. This has real implications for the liberties of New Yorkers

Case in point: license plate readers sharing their huge repository of data and movements with ICE.

Self-driving vehicles, like facial recognition systems generally, perform poorly when analyzing communities of color. Without oversight, Kaufman rightfully says we're inviting yet another risk: car accidents.

Janai Nelson from the NAACP LDF calls on the task force to recommend full transparency of the NYPD's use of ADS. The potential impacts on communities of color, combined with the department's history of unconstitutional policing makes any carveouts for NYPD unacceptable.👏🏽👏🏽👏🏽

Publicly identify, categorize, and share a list of the NYPD's use of surveillance technology. This list should be continuously updated.

The #POSTAct requires this!

Nelson walking through many of the points made in Dirty Data, Bad Predictions: How Civil Rights Violations Impact Police Data, Predictive Policing Systems, and Justice: papers.ssrn.com/sol3/papers.cf…

Nelson gave a brief but damning overview of NYPD's unconstitutional and discriminatory practices. Any ADS that relies on this historical crime data will be irreparably tainted.

NAACP LDF calls for an outright ban on any ADS that relies on historical NYPD crime data.

No sugarcoating: ADS threatens to completely redefine everything ranging from reasonable suspicion to freedom of speech.

Task force member @mer__edith asking about how to evaluate systems for bias before they're deployed on public data. This is a real concern, and the panelists aren't convinced existing audit oversight mechanisms can do the job.

An ongoing topic of conversation: impacted individuals. There are great advocates here, but making this forum at 6PM on a weeknight inside a Tribeca law school is not the way to meaningfully engage the public at large.

The topic of ADS doesn't need to be esoteric or confusing. Any member of the public can understand why its scary to have computers making decisions about their education, housing, and freedom.

Another good question: who should determine when a data set is too tainted to use in an ADS? Harris recommends it should be a 3rd party such as a city’s CTO or Chief Analytics Officer. This could put a lot of power in a Mayor’s hands...