Three takeaways from my Grace Hopper Conference experience
By Jinlin He
What does it feel like to see inspiring female engineers who share the same childhood dreams as you? What does it feel like to meet female engineers from all walks of life who completely defy your definition of what an engineer should look like? What does it feel like to be in a tech community where everyone wants to support you and see you succeed? One week ago, at the Grace Hopper Conference (GHC), like every female engineer I met, I’ve never felt so right to have chosen technology as my way of empowering lives.
I felt so right to be in the field of technology not because it is new and exciting, but because it is “tricky.” Yes, we are in an age where new technology emerges at an exponential rate, but how it can be used requires so much art and creativity, and how it should be used demands empathy and integrity. What Padmasree Warrior, CEO & Chief Development Officer of NIO said in her opening keynote speech really stuck with me, “What does it mean to be a leader at an age of such rapid change and uncertainty? You have to offer assurance. You have to be authentic.” Therefore, the diversity in tech we have been talking about goes beyond demographic markers. It is about respect for individuality and non-conformity. When Jessica O. Matthews, founder and CEO of Uncharted Power did a demo of her patented electricity-generating jumping ropes in her bright red high heels, I saw a confident female engineering leader. But more importantly, I saw a bold female engineering leader who started with one single idea — let the community solve their energy dependency problems by playing sports that they are already so passionate about — and went for it. This is the spirit and promise of technology. An open environment that celebrates difference and authenticity is what we need to create for our next generation of technology talents.
To put the above points into perspectives, I want to highlight these sections from GHC:
ACM Award Winning Research in AI
In this session, Jieyu Zhao presented her paper Men Also Like Shopping: Reducing Gender Bias Amplification Using Corpus-level Constraints. The presentation highlighted astonishing results of how machine learning algorithms predicted the gender of a person who is cooking in a photo as female while every feature of the person’s appearance shows that he’s male. This implies that the algorithm has learned to associate the activity he is doing (cooking) heavily with female.
Another example illustrated in the slide above is coreference resolution in natural language processing. Jieyu presented an algorithm that easily linked the term “president” to the male pronoun that appear in the subsequent sentences. However, when the example changed “his” to “her” while keeping all other words the same, the algorithm did not make such association any more.
So what is the problem? The speaker pointed out one of main reasons is the imbalance in training data set where photos of female cooking appears 33% more than the male counterpart. In addition, what is surprising is that such difference is significantly amplified through the algorithm. This is because what the machine learning algorithm is doing is making a prediction that appears like learned data. Given our current data that inherently captures our existing bias, it is bound to echo, and in this case, amplify such bias. The takeaway for all machine learning practitioners out there is that data is not objective, but subjective. It serves as a mirror of our biased society.
Using Computational Creativity to Create Music
Anna Chaney, a data scientist from IBM, shared how the computer system IBM Watson uses reinforcement learning and a deep belief network to creatively generate melody and beats. Only referring to western music theories as a base line specification, the model can take full advantage of the concept of “exploration” and compose music in a creative manner.
This is a really interesting aspect of AI. What does creativity mean for AI? Form the answer to this question, we can learn more about what creativity means for humankind. Indeed, now is the time when different domains start to merge rapidly. Yet there is one direction that all technology is heading — people. Thus technology leaders always need to have a great understanding of humanity before they understand technology.
Below is a clip from an album that’s collaborated with artist Taryn Southern:
Celebrating Chinese Women in Technical Roles
Julia Liuson shared her journey from a software engineer to an executive at Microsoft. Though she addressed the session specifically to Chinese women in technology, the advice applies to any engineer leaders who aspire to transition from technical role to a management role. She advised a change of mindset: in addition to creating values through one’s own work, think about how to empower others and add values to the team. Julia said that she gained more understanding of management after she became a parent. Just like any parent would willingly make the best decision for their kids and wish their kids to be better than themselves, managers should share the same mindset of making the best decision and creating the safest environment for the team.
This is particularly relevant to my own learning experience at the early stage of my Capstone project. To always begin with the thought, “How can I empower collective learning?” is a novel process for me in a professional setting.
The three days at the Grace Hopper Conference gave me a lot of new perspectives that I cannot highlight one by one. As female engineers from different companies with different social and professional backgrounds came together at Houston, Texas, we are re-centering ourselves as socially conscious technology builders. In fact, we have to — particularly in this exciting yet chaotic time of technology.
Jinlin He is a UC Berkeley Master of Engineering candidate in Industrial Engineering & Operations Research, Class of 2019. Connect with Jinlin.
