top of page

The Importance of Ethics in the Development of AI

Updated: May 14, 2019

The Data Lab, an Edinburgh based research lab, last week hosted DataLab 2018. This year the conference focused on the implications that AI has for society and the important role that ethics will play in the development of such. One of the speakers, Megan Hughes the Artificial Intelligence & Social Media Analytics Lead at Accenture, broke down the concepts of AI development into the three myths, three imperatives, and three challenges that the field faces now and in the future.

Starting with the myths, Megan explained the fears that the community as a whole has around AI. The first myth, clearly influenced by popular sci-fi movies, is the classic “robots are coming for us” idea. This creates the idea that AI development is essentially an “Us (humans) vs. Them (AI)” scenario, which Megan quickly dispelled by explaining that data scientists believe it to instead a be beneficial relationship. The second myth covered a fear closer to home, that of job security. Many people believe that AI will quickly make many jobs obsolete, leaving multitudes unemployed. Megan did not deny the fact that AI will take over certain jobs, but followed with explaining that for every job AI eliminates, another IT job is created. It is recognized that AI will have negative impacts on the labour market, but according to some statistics, there will soon be a labor shortage in the realm of IT. The final myth was that the current static approaches to AI will still apply in the future applications. In other words, the same methods used in the early days of GPS will still be used in the developing apps, which of course can already be proven false by an app as simple as WAYZ, a dynamic GPS system. This is indeed one of the more detrimental myths, because if a company remains stuck in its old ways, it will be quickly left behind in the fast-paced development of the field.

Following the myths, Megan touched on the three imperatives that AI is bringing to the work force. The first two were the facts that AI is allowing people to reimagine business as well as a new approach to work. All one needs to do is take a look at the Silicon Valley and the companies coming from that startup driven culture to see the strong evidence behind both of these claims. It was Megan’s third imperative that really caught the attention of the data scientists scattered throughout the crowd, and that was the fact that we need to start focusing on responsible AI. This essentially means that the AI being developed needs to have a “human” center, minus the bias that seems to be inherent in human nature. There must be a sense of trust, transparency, honesty, and justice embedded into these new AI systems if this technology is going to advance in an ethical manner.

To finish her talk, Megan covered the three challenges AI now faces in moving forward, the first of which was the lack of skillset in the field. There is a difference in understanding between those who use AI and those who work with it, and currently there is a growing gap in skillset for those who are working with AI. The next challenge dealt with data veracity, as AI needs data to function but not just any data, the correct data. People need to understand the source of the data as well as the context in which it was collected to ensure that the data is not corrupt. And the final challenge was the fact that there really is no finish line with what AI development can do. The problem is not how far can we take AI, but rather how far should we take AI, and right now no one has bothered to put any limits in place.

Although the myths, imperatives, and challenges covered a wide range of facts and ideas, the common thread among them all was the call for an ethical standard and practice. Gone are the days of AI development just for the sake of development. The conversation is changing, and ethics is quickly coming into necessary focus.

What does this mean for Silicon Valley companies?

The definition of success is no longer whether the AI system works or not, it is if it works according to an ethical standard. Foreshadowed by the GDPR, ethical practices are quickly becoming essential to the business process.

In fact, the central issue of the upcoming data privacy conference in Brussels is ethics and privacy of data. We will make sure to gather all evidence and arguments that there is.



bottom of page