Last year, Michigan's Integrated Data Automated System flagged more than 540,000 unemployment claims as possibly fraudulent. Thousands of state residents’ accounts were inaccurately flagged, making it almost impossible to re-establish their legitimate claims.
Closer to home, the Ontario Human Rights Commission recently identified serious human rights concerns about the “potential discriminatory impact of police data collection” with predictive policing algorithms and facial recognition.
Ethical AI is going to be central to May Masoud’s keynote at the upcoming Vancouver DataJam. SAS' data science solution specialist and classical statistics and modern machine learning expert is one of the keynote speakers at the virtual one-week, hackathon-like event.
“The conversation isn’t about what AI can do for us, but how AI can be more responsible,” she said.
Masoud will cover how organizations can build a framework to hold themselves and their AI systems accountable.
When an automated decision is made, such as a denied loan application at a bank, can the process behind that decision be explained? It's an important question, but only one of many on the path to accountability. The initial collection of data plays a major part in this discussion as well.
At SAS, Masoud helps enterprises derive a measurable return on AI investments while maintaining governance and ethical considerations. She says she wants to prevent the term “ethical AI” from becoming another buzzword.
“It’s a big conversation, but not enough companies are taking action on this,” Masoud said.
Ethical AI and its role in COVID recovery
This year’s DataJam event is focused on COVID recovery. Participants are working on data-related projects centered on the following: social recovery, sustainable recovery, economic recovery, and public health recovery.
Topics such as vaccine rates, shopping behaviours, and access to resources are just some of the areas to be explored.
The free social coding event brings together Canada’s young programmers and data scientists, and this year, also features three keynote speakers, including Masoud. It's also entirely volunteer-run.
“We’re a grassroots organization, so it can be difficult for us to put events like this together and we’re grateful for how SAS has responded," said Laura G. Funderburk, chairwoman of the DataJam.
Funderburk says the event has quickly fostered a community of leaders who understand the importance of diversity in the workplace.
"We have a strong focus on gender and racial minorities. We want to facilitate connections, provide networking opportunities and recognize diverse voices in tech and data sciences," Funderburk explained.
Masoud noted how legislation around emerging technology progresses at a snail's pace. Events like the DataJam help initiate important discussions while legislation catches up.
For example, the Canadian federal government's proposed Consumer Privacy Protection Act (also known as Bill C-11) remains in limbo. C-11 has been stuck in second reading since its initial introduction last November. Bill C-11 is meant to bring Canada's federal privacy laws in line with the European Union General Data Protection Regulation.
Until then, organizations using personal information without consent can sidestep accountability with relative ease. The problem extends to Bill C-11 itself. Canada's Privacy Commissioner Daniel Therrien recently stated that C-11's administrative penalties will not apply to the most important violations relevant to consent.
"Don't wait for these laws to come into effect," Masoud stressed. "Do it by design."