As this article bluntly puts it, “Artificial Intelligence has a ‘sea of dudes’ problem.” According to Hanna Wallach, AI researcher and cofounder of the Women in Machine Learning Conference, only 13.5% of those working in machine learning are female. At Kiite, our data tells a different story. Our current team is 64% percent women, including two executives, and one board member.
Artificial intelligence is still at inception. We haven’t yet seen its full innovative potential, or how it will change our everyday lives, yet AI is already embedded in our surroundings and is making our lives easier — our homes, phones, thermostats and cars are already self-learning. As we become increasingly dependent on these technologies we must ensure that we are building them responsibly.
It’s inevitable that AI will reflect the values of its creators, which is why inclusivity is essential in all aspects of developing such impactful technology. From how we develop our algorithms, to constructing and curating data sets, if we aren’t thoughtful and deliberate, we’ll perpetuate ingrained forms of bias and harmful stereotypes indefinitely.
It comes as no surprise that training data is one of the hottest and most controversial topics in machine learning. Training data often contains real world information which is fed to computers and eventually reflected back to us through AI technologies, like Siri’s responses to our pressing questions. Without paying proper attention to fairness, accountability, and transparency of what we’re building, intelligent machines could inherit our biased viewpoints based on questions or data. That’s why gender diversity matters. In a world where STEM programs are male dominated, we must be deliberately inclusive.
We believe that the ones building the technology should be the ones setting the standard.
In recent years there have already been numerous examples of algorithmic flaws containing bias within AI technology — some of which we discuss in our latest article, “With great data comes great responsibility and why products like Faception can be dangerous”. But like with all things, to get anywhere, you have to start somewhere.
The truth is, algorithmic flaws aren’t easily discoverable. Not by the public, or even the technologists building the products. For example,”[i]n 2015, The Ad Fisher team found that when Google presumed users to be male job seekers, they were much more likely to be shown ads for high-paying executive jobs. Google showed the ads 1,852 times to the male group — but just 318 times to the female group.” Without the Ad Fisher team bringing these insights forward, would the public have realized what was happening? It’s doubtful.
Fortunately, the push for more diversity and inclusion in the ICT tector is an ongoing conversation. Awareness is spreading and companies are making moves and introducing programs and policies that build a diverse hiring pipeline. In the local Waterloo tech community, executives and companies are speaking out, like Vidyard’s CEO, Michael Litt, who’s recent article, “I accidentally build a brogrammer culture and now we’re undoing it,” is encouraging other leaders to step up and create change.
In celebration of International Women’s Day, we want to highlight some of the women on our team who directly handle labelling, classification, extraction, and training of our data and AI technologies at Kiite.
Aycha Tammour, Data Scientist
Aycha, a Data Scientist at Kiite, uses machine learning algorithms to examine large amounts of text data and analyze their contents. ML algorithms allow us to uncover the contents of new documents based on our analysis of previous similar documents. Aycha looks into things like the grammatical structure of a sentence and how often a given term appears in a document to help us build mathematical models of the text. These models are then capable of capturing important aspects of the text document such as topics or the sentiment in its various passages.
Melody Zapotoczny, Linguistic Analyst
A polyglot and language enthusiast, Melody is Kiite’s Linguistic Analyst. She works closely with our implementation and research teams to ensure that the data we’re using to train our models properly reflect real-world scenarios and that our training models meet our internal research requirements.
Chengpei, Senior Software Developer
Chengpei is on the engineering team at Kiite and through her role as a Senior Software Developer, she works on building the Kiite application, which customers use to access analytic reports, manage their knowledge base and configure software settings. On a daily basis, she works with modern tools such as React, Redux, and Node among others. Chengpei also works on enhancing Kiite’s REST API and bot functionality using microservices.
Koren, Implementation Lead
Koren, Kiite’s Implementation Lead, does it all. She provides ongoing training and guidance to our other Implementation Team team members, and she ensures that our customers are well taken care of and that their data is securely handled. She helps to make our customers successful at every stage in their life cycle. That means Koren not only develops data migration strategies and executes on them, she also runs live user training sessions to ensure that our customers know how to use Kiite!
Kay Lazarte, Implementation Specialist
ay is an experienced business process professional and a key part of the Implementation Team member at Kiite. She helps gather customer requirements and works on projects to help us develop our data models. She also has a passion for security and makes sure that projects follow data security and privacy standards at all times.
Donna, VP of Operations
As head of operations at Kiite, Donna’s responsible for making sure that our work environment supports our company’s growth and momentum. This includes ensuring that we’re supporting best-in-class data management practices. She works with our implementation and research teams to ensure that our data pipeline is meeting both internal and external customer requirements. She knows that customer happiness is key and believes that the best way to make customers happy is to build the right team for the job.
