Open mozfest-bot opened 6 years ago
jontutcher
commented
6 years ago Hey @Nolski - nice topic & submission. Would you be up for collaborating with others in the same area on a session (either as part of this one or separately) if anyone else is about?
Nolski
commented
6 years ago @jontutcher I be very much interested :) Definitely my main goal with coming to MozFest would be to collaborate with others in the space so perhaps having a discussion panel or something would be super interesting to compare AI ethics across multiple spaces.
Any idea of other talks in the same space that we could combine?
KristineGrivcova
commented
6 years ago Hi @Nolski ! Would the session include some kind of interactive part in addition to the presentation? Thanks, Kristine
Nolski
commented
6 years ago Hi Kristine!
So what I was thinking is that it will start off as a presentation for perhaps 15 minutes to lay the concepts and current work that has been done so far to give the audience a good base to begin discussion. I hope after that, my role as the presenter will be to foster discussion among audience members.
I plan to do that by employing a few strategies:
I will bring with me a large set of anecdotes and statistics to help bring some structure to the presentation. This will hopefully allow the whole session to be very non-linear and driven by the audience.
If the audience is huge, I will likely split up the audience into groups and do kind of co-design sessions where a certain context is introduced, and individual groups will analyze the use of AI in that context, and present on their findings and risks (this is inspired from the Digital Privacy Impact Assessment that you often see done for GDPR compliant projects).
It will be very discussion focused but I hope to make it a bit more entertaining with visuals, hopefully fun imaginative scenarios and interesting anecdotes from my time working in the humanitarian relief sector.
Let me know what you think of the concept! Even if it doesn't get accepted for the conference, I'm always happy to hear feedback :)
[ UUID ] ab2ea486-53cb-47f5-b794-00dcfaa57c28
[ Session Name ] Artificial Empathy: Ensuring the Machines Better Humanity [ Primary Space ] Digital Inclusion [ Secondary Space ] Privacy and Security
[ Submitter's Name ] Mike Nolan [ Submitter's Affiliated Organisation ] International Rescue Committee [ Submitter's GitHub ] @nolski
What will happen in your session?
As our ability to collect and analyze large swaths of data has increased dramatically so too has our regulatory capacity. With the introduction of regulation such as GDPR and DPIA, organizations are forced to consider their collection of data in terms of ethical consequence.
As the popularity of machine learning algorithms grow, as will their impact on livelihoods. In the case of law, humanitarian aid, and many other circumstances, the outcome of these models can greatly affect the lives of the users they touch.
In this session, we will explore the biggest challenges we face with machine learning in the humanitarian context and what it will take to define the guidelines for building an ethical classification algorithm.
What is the goal or outcome of your session?
Our goal for this session is to create the building blocks through a participatory discussion that will serve as an open guide and scaffolding for considering and evaluating the ethical implications that a classification algorithm may have upon beneficiaries.
We will draw from past experience and use existing projects provided by the speaker as a metaphorical to build and test our guidelines upon.
If your session requires additional materials or electronic equipment, please outline your needs.
A projector would be great in order to present a slide deck.
Time needed
60 mins