DAIN Studios attended the Data Natives Conference, which is the meeting point for data-minded industry experts, entrepreneurs, tech and business professionals to inspire each other and disrupt the status quo.
The Data Natives conference is an excellent industry pulse check on the discourse about Artificial Intelligence (AI) and Data Science in Europe. This year, it was noticeable that the term AI is no longer a magical marketing hook, because people from a variety of backgrounds are gaining a deeper understanding of this technology and are actively joining discussions. With a much larger following, AI came under more scrutiny and its limitations were highlighted, with many seeing it as a technology that complements human intelligence rather than replacing it. All in all, there were many exciting talks and lessons learnt during these 2 days in Berlin, however 3 themes dominated the Data Natives conference: explainability, overcoming demographic bias and using data for good.
Explainability is one of the hottest topics in the AI and Data community this year and has been featured as one of the peaking technologies in Gartner’s 2019 Hype Cycle. There is general consensus across most users, that the black box models cannot be used when they impact humans, and that we need to interpret the outcomes and reasons behind them. Popular tools like Local Interpretable Model-agnostic Explanations (LIME) and SHAP (SHapley Additive exPlanations) are recognized to be great for explaining computer vision model results. But computer vision is only one area of AI, Uri Goren of Nym has argued that in the case of Natural Language Processing, it is best to use a hybrid approach that combines Deep Learning with traditional computational linguistics practices. Overall, we saw a trend of shifting from pure machine learning applications to incorporating deep domain knowledge.
Overcoming Bias. A significant number of the talks and panels were dedicated to biases found in the current AI models. At this stage, the majority acknowledges this unfortunate phenomena and is ready to work on solutions. For example, Gunay Kazimzade, a TU Berlin researcher who focuses on gender and racial bias in AI, proposed an action plan for data practitioners to avoid common pitfalls and minimize the negative impact of data science and artificial intelligence products, which includes looking at the quality of data, building awareness about biases and taking into consideration different sections of society ahead of starting the project.
Creating positive impact. Perhaps, the most overbooked section of the conference was the newly added “Impact track” that featured speakers discussing the possibility of shaping a brighter and more just future with the help of data. The conversations revealed that there is a vast amount of Open Data that could be used for public good and, reversely, the public would benefit greatly if private companies shared even a fraction of their data to help civil society better organise cities and public services. There is a clear inclination for collaboration between the data experts, hungry to switch up their roles in e-commerce and finance businesses to something purpose-driven, and the not-for-profit sector that is lagging behind in digital capacities.
Following scandals around unfair algorithms and data-hoarding tech giants gaining perhaps way too much power in the recent past, society has seen many misinformed assumptions and disparities within the discourse of data and AI. Yet, the demand and active participation in the talks at the Data Natives 2019 indicates that the civil society and data practitioners are still hopeful about using data for the greater good. We are expecting to see more public-private cooperation in this area as well as brand new social and ecological data-driven initiatives coming out in the near future.