This week, I’m happy to review a couple of areas where I feel data science is making important moral advances. First of all, the idea of paying users for using their data; and secondly how social networks are testing ways to reduce the pressure of posting to their platforms.
As A.I. continues to influence more and more of our lives, it’s increasingly important to pay attention to the implications – not just the profits or perceived conveniences.
Paying Users for Using Their Data
In last week’s Tech Tuesday about 5 ways to improve business with data science, I shared the article: Facebook is trying to make A.I. fairer by paying people to give it data by Rachel Metz for CNN Business (link).
In a push to make A.I. less biased, Facebook created a dataset of diverse individuals they paid for participation. They hope researchers will use the open-source data set to help judge whether A.I. systems work well for people of different ages, genders, skin tones, and in different types of lighting. Research participants were compensated for their participation and can remove themselves from the dataset at any time for any reason.
Paying participants who help scientists train A.I. has the added benefit of increased accuracy, as the participants are able to accurately report their demographics and gender.
The hope and intent are that this type of labeling will lead to fairer A.I. systems, both in terms of predictions about individuals as well as accountability in the training data, and that we will come to see more positive actions like this from all the biggest players in A.I.
Reducing the Pressure to Post
Next up, regarding how the “likes” that social posts gain affect us all, Adam Mosseri, head of Instagram, recently tweeted the following:
Last year we started hiding like counts for a small group of people to see if it lessens some pressure when posting to Instagram. Some found this helpful and some still wanted to see like counts, in particular to track what’s popular.
So we’re testing a new option that lets you decide the experience that’s best for you – whether that’s choosing not to see like counts on anyone else’s posts, turning them off for your own posts, or keeping the original experience.
We’re testing this on Instagram to start and we’re exploring a similar experience for Facebook too. More to share about this soon.
This is a very important move in the right direction.
Striving to create posts with more “likes” may cause increased anxiety and lead to depression and a never-ending desire to impress others that starts all over again with each new post. This constant comparison may not be healthy, especially to adolescents and especially to females. This is particularly true for those who already suffer from feelings of depression and anxiety.
Allowing users to hide likes likely won’t have much impact on revenue, and it remains to be seen whether it is actually used by those who may benefit from it most.
I would love to hear your thoughts on this and will keep following these topics as they progress.
Have a good week,
-Zank
Of Interest
High-Res 3D Models of Humans Created with a Single Photo
Recent advances in image-based 3D human shape estimation have been driven by the significant improvement in representation power afforded by deep neural networks. Although current approaches demonstrate the potential in real-world settings, they still fail to produce reconstructions with the level of detail often present in the input images. The videos are fascinating, you can find them here.
Pragmatic and Fair A.I. in the Real World
In this article data scientist David Graus contends that A.I. doomsday scenarios are slowly being superseded by views of A.I. in which we are in control of A.I. and able to design them in ways that reflect values of our choice. The message, he writes, is clear: we can and should operationalize algorithmic fairness. And to do this, to build useful and fair A.I. tools, he argues that we do not first need to “solve” the trolley problem before allowing self-driving cars, or to fundamentally eradicate injustice, or rid ourselves of bias entirely. Read his article on thoughtfully (co-)designing A.I. systems here.
Twitter Outlines New Approach to Algorithms
Algorithms have become a source of rising concern in recent times, with the goals of platforms feeding us information often being at odds with broader societal aims of increased connection and community. Various studies have found that what sparks more engagement online is content that triggers strong emotional response, with anger, for one, being a powerful driver of such. Given this, algorithms, whether intentionally or not, are often basically built to fuel division to maximize engagement. Various platforms are now examining this and Twitter has outlined its latest algorithmic research effort, the ‘Responsible Machine Learning Initiative‘.