Data Ethics

The progress of digital disruption and transformation has increased consumer concerns about the way data are collected and utilized as more and more transactions move online. To meet these concerns, organizations will need to reassure their customers that data privacy is an essential part of their business model. Failure to do so could result in pushback and increased legal challenges. 

The first step is the development of ethics guidelines. One example is the U.K. government's Data Ethics Framework, designed to help teams plan, implement and evaluate new projects. Below are four considerations when drawing up guidelines: 

1. Go beyond the letter of the law. “The law is always trying to play catch-up with technology, and corporations sometimes try to skirt around the law,” says Wanda Curlee, program director, business administration at American Public University System. Teams should set objectives and follow best practices, along with creating a framework that encompasses future laws. 
2. Get a diversity of views. Clarifying concepts like privacy, fairness and transparency, and what they mean for everyday practices across cultures, isn't easy. Ethics managers need diverse teams, and input from a wide range of stakeholders—including project managers—as part of ethics committees and more informal interactions.  
3. Consider all impacts on all stakeholders. The needs of direct and indirect stakeholders—and intended benefits and unintended consequences—must be assessed and communicated, with a consideration of solutions and trade-offs. 
4. Continuously review guidelines. Curlee says ethics guidelines “need to be reviewed constantly because technology is changing constantly.” And what's acceptable today might not be tomorrow. 

Guidelines have little use if they aren't integrated into project planning, practices, communications, metrics and reporting. Ethics training will help, as will having an ethics representative on projects. There must also be ways to report breaches, openly or anonymously. The sheer amount of data generated on big projects will make algorithmic audits necessary, but Curlee cautions that we must understand the algorithm that's helping do the audit. 

Wanda Curlee is program director, business administration at American Public University System. She conducts research in artificial intelligence (AI) and has been a project manager for over 30 years. 

PMI spoke with Wanda Curlee about a range of data ethics issues. Here, she shares her views on algorithmic fairness and control.  

“The chief data or ethics officer has the moral responsibility to understand what's going on in that black box. If they do not understand what's going on in that black box, then that black box should be turned off,” Curlee says, referring to any data analytics system for which we observe inputs and outputs but not the internal workings or algorithms. “It’s learning all the time, and if it's learning incorrectly, you're going to have biased or unethical or wrong data. Period.” 

But, Curlee emphasizes, “Everybody is responsible, from the person who designs the algorithm to those who use it. If I'm not getting correct answers or the answers appear to be biased, then I have an obligation to speak up. People need to understand and constantly test the black box.” 

Ethics considerations need to extend throughout the IT supply chain, says Curlee. “When you're buying that technology, you have to understand the biases and you need to correct for it. When you retire or sunset that software, what are you going do with all that data?” 

Finally, Curlee advises project managers to overcome any fear they may have of AI—instead, approach AI and other disruptive technologies as an opportunity to differentiate themselves. “I just want project managers to be excited about AI and learn about AI. Because if they do, and they know the questions to ask and they know how to use AI or any disruptive technology on their project, they are going to be more valuable than that project manager who doesn't.”