Machine learning is now being used to make crucial decisions about people's lives. For nearly all of these decisions there is a risk that individuals of a certain race, gender, sexual orientation, or any other subpopulation are unfairly discriminated against. Our recent method has demonstrated how to use techniques from counterfactual inference to make predictions fair across different subpopulations. This method requires that one provides the causal model that generated the data at hand. In general, validating all causal implications of the model is not possible without further assumptions.

This paper tackles the primary criticism aimed at applications of causal graphical models for fairness: one needs to completely believe an assumed causal model for the results to be valid. Instead, it presents a definition of fairness where we can assume many plausible causal models and requires fairness violations to be bounded below a threshold for all such plausible models. The authors present a simple way to formally express this idea: by defining an approximate notion of counterfactual fairness and using the amount of fairness violation as a regularizer for a supervised learner. This is an important theoretical advance and I think can lead to promising work. The key part, then, is to develop a method to construct counterfactual estimates. This is a hard problem because even for a single causal model, there might be unknown and unobserved confounders that affect relationships between observed variables.

Machine learning is now being used to make crucial decisions about people's lives. For nearly all of these decisions there is a risk that individuals of a certain race, gender, sexual orientation, or any other subpopulation are unfairly discriminated against. Our recent method has demonstrated how to use techniques from counterfactual inference to make predictions fair across different subpopulations. This method requires that one provides the causal model that generated the data at hand. In general, validating all causal implications of the model is not possible without further assumptions.

There's something big afoot in Fortnite: Battle Royale, and Epic wants all eyes forward. The developer just updated the in-game news feed with a somewhat cryptic message advertising the upcoming Season 5 battle pass, something that dedicated players have been speculating about for a while but that Epic has only recently started to address head-on. The text reads only "Worlds Collide: 3 days until Season 5." The message is accompanied by a teaser image that Epic's official Fortnite twitter sent out a little while ago. It appears to depict a stylized version of a Japanese mask depicting Kitsune, which is both the word for fox and a sort of character out of folklore and mythology. There's also a blue, crackling rift over the right eye, something that matches what players have been seeing in-game for about a week.

The internet has evolved a great deal since it was first launched back in the 80s. Back then, there was no such thing as a high-speed internet connection, hence the need for simplistic, text-based website with no variations in font, typography, layout or visual content. A website back then consisted of nothing but pages and pages of standard text placed in front of a white background, with no pictures and no font variations except those in size. One could argue that reading a book was probably a more enriched experience back then. Fast forward a decade, and we got our standard wired connections.

What I am saying applies specifically to Grid, the "revolutionary" startup company that was the first to announce that it has created a website builder that uses artificial intelligence to design the perfect web destination. Reviews, however, have been rather mixed, not quite perfect as they promised. The involvement of the user is minimal, which, on the plus side, goes to show that whatever the end website looks like is entirely credited to the AI itself and not the person pressing the buttons. That, however, also means that the user has less than enough freedom to customize their website's layout as per their wishes, even if they are a veteran in the subject. The AI limits the website's customizability to a few choices of layout and color only, asking the user to accept the fact that the algorithm knows best.

Missy Cummings gets recruited for jobs outside of academia a lot. But she always says no. Cummings, a professor in the department of mechanical engineering and material science (aka robotics) at Duke University, has consulted for many autonomous drive programs for automakers and Silicon Valley companies, but she said it's better for everyone if she just stays where she is for her full-time gig. "I've had many offers to leave academia, but I'm doing everyone a favor by staying in academia," she said. "I am very persnickety about data and evidence and I would drive these companies crazy." As companies such as Google race to bring autonomous cars to market, the cultural divide between academic types and corporate types is surfacing publicly.