Capital One artificial intelligence (AI) researcher Omar Flórez knows what it’s like to be an outsider. In AI, where English is the lingua franca, achieving success can be a challenge for those who aren’t fluent. Omar grew up in Peru speaking Spanish. When he started learning about AI, it was difficult to find technical documentation in Spanish. And when he traveled abroad to pursue his interest in the field, at times it was difficult.
“I had to translate everything I heard in my mind,” he says. “It was the only way I knew to make sure I didn’t miss anything important.”
Omar observed that for people around the world, whose first language is not English, learning Artificial Intelligence is challenging because it requires them to translate every single term into their native language, sometimes with no intuitive meaning. Omar observed that this could be a problem in the technology industry with significant costs: missed opportunities for hiring diverse talent, technologies that are not optimized for all customers, and implicit cultural bias.
The answer, he says, is to ensure that everyone everywhere enjoys the same access to the content necessary to contribute to the technologies that will shape the world of tomorrow.
“When it comes to achieving diversity in technology, it all boils down to education,” he says. “The playing field isn’t level if people don’t have equal access to knowledge, and as a result technology doesn’t serve everybody as well as it might.”
Driving innovation with diversity
Omar joined Capital One in part because it placed him at the cutting edge of AI within a company that had immense, compelling sets of data opportunities. In his role, he conducts state-of-the-art research with the goal of enabling natural conversations between humans and machines, especially with Capital One’s intelligent virtual assistant, Eno. Currently, he focuses on using meta-learning and memory-augmented neural networks to learn with small data, reuse knowledge bases, and manage catastrophic forgetting in neural network models.
“In my work, I’m trying to make it as natural and fluid for you to talk to a machine as it is to have a conversation with me,” he says. “One of our recent breakthroughs was discovering that certain new types of neural networks actually work more efficiently with less data—which has amazing implications for enhancing natural conversation.”
But it was more than just technology that convinced him Capital One was the right place for him. The company has a genuine commitment to diversity, he has observed. It recruits candidates from varied backgrounds, maintains active groups and communities so everyone gets the acceptance and support they need to thrive, and backs employee participation in diversity efforts in the broader professional community, such as LatinXinAI, a community focused on furthering innovation and access to resources in AI for LatinX people across the world.
When it comes to bringing diversity-enabling AI ideas to life, Omar isn’t just talk. He created the Spanish-English Dictionary of Technical Terms in Artificial Intelligence (AI), a first-of-its-kind resource in the industry, and found encouragement and support from his colleagues. Capital One sponsored the project’s presentation at a conference for diversity in AI and hosts it in the company’s public GitHub repository. The dictionary has been well received since its release; today, it helps those who want to write algoritmos de aprendizaje profundo get a handle on AI lingo much faster than before.
Creating a more equitable future
Omar’s vision for the future is one where there are no outsiders, and where companies and organizations serve all people equally via democratized, bias-free technology. There’s still a long way to go, but he’s confident AI can help create a better future for Capital One customers, and, ultimately, the entire financial services industry.
“We’re in the early days of being able to fully remove biases from algorithmically generated content,” he says. “But I’m glad that at Capital One, the responsible development and deployment of this technology is a priority across the company -- we are all in on doing AI the right way, which is going to mean better banking for everyone.”
Omar believes AI will empower customers like never before, giving them automated advice tailored to their unique financial situation and coaching that anticipates their needs.
“We have the right talent and technology and massive amounts of data to train our algorithms,” he says. “It’s going to be an interesting journey, and I think the work we’re doing on language is going to play a central role in getting us where we want to be.”
- Favorite hobbies: Travel, SCUBA diving
- Fact about you that people are most surprised to learn: “I dance salsa very well.”
- Currently reading: The Future of the Mind by Michio Kaku
- Historical person you’d like to meet: Nicolaus Copernicus