algorithms

UC Berkeley’s Anca Dragan says algorithms need to get better at understanding human needs and behaviors.

Tijmen Schep coins a phrase to describe the chilling effect of the reputation economy on the expression and exploration of ideas.

Mike Loukides explains why predictive algorithms must be imperfect (and error prone) in order to contend with real-world data.

Alexis Madrigal considers the creative future of machine-generated sketches—and what it teaches us about how humans think, too.

Joy Buolamwini surfaces bias in facial recognition software—and responds to some screwed-up criticism.

Bloomberg peeks in on the tedious worklife of humans pretending to be robots pretending to be humans.

Facebook is prototyping its omniscient bot by powering it with people, the MIT Technology Review reports.

In Backchannel, David Weinberger considers the implications of machines constructing their own models for understanding data, quite divorced from our own (more simplistic) models.

MIT Technology Review covers the unknowable logic of our most sophisticated algorithms—and the implications of having no idea how they actually work.

Google’s move to highlight fact-checking sites may be a promising step in identifying hostile information zones in search results.

Memories of a night in Siena remind me what we’ve gained—and what we’ve lost—in the era of the algorithm.