6 min read
Feb. 5, 2024
Responsible AI for the greater good: insights from AWS’s Diya Wynn
Diya Wynn, responsible artificial intelligence (AI) lead at Amazon Web Services (AWS), reflects on the continuous evolution of an emerging technology and the opportunities it will create for the future
Written by the Life at AWS team
Diya Wynn
Responsible AI Lead at AWS
Diya Wynn, responsible AI lead at AWS, was thinking about how to prepare and educate her two school-aged sons for the future workforce when, around 2019, she said there were three main trends dominating conversations around the future: big data and the move toward data-driven everything; artificial intelligence (AI) and robotics; and virtual/augmented realities.
As she explored these topics further, she started recognizing an absence of diverse voices shaping these trends.
“That absence can result in systems of inequity,” Wynn said. “I’ve always been a change agent and thought there was an opportunity to have an impact.”
As an AWS technologist who was focused on cloud transformation projects at the time, Wynn started thinking of ways AWS could guide its customers responsibly when they use AI. She envisioned an operating approach that would encompass people, process, and technology in order to minimize risks and unintended impacts while maximizing benefits to business, customers, and society as a whole.
“In Amazon’s culture, anybody in any role can think big, have an idea, and present that idea to leadership in the way of a narrative or proposal,” Wynn said. “I knew we could have an impact to help influence the hundreds of thousands of customers using our AI services to help them be more conscious, intentional, and build in ways that are inclusive and fair. A few years ago, my proposal for a customer-facing practice focused on responsible AI was approved and funded.”
Fast-forward to the present after hundreds of conversations with customers, partners, scientists and academia, Wynn reflects on a journey that continues to evolve daily thanks to the pace of innovation and growing public interest around AI and ML (machine learning) technologies.
“AI is believed to be the most transformative technology of our time. It is being used every day across nearly every industry globally. It is changing the way we live, work, play, and engage,” she said. “And AWS has more than 100,000 customers using our AI/ML services to build these systems that touches tens of millions of users around the world. My goal is to change the world, one AI at a time, by influencing how people are designing, building, implementing, and using AI.”
“AI is believed to be the most transformative technology of our time. It is being used every day across nearly every industry globally. It is changing the way we live, work, play, and engage."
People and guardrails
Wynn works with customers to ensure that when they build AI and ML systems, they take a people-centric approach. Since AI/ML systems are designed according to the specifications given, AI/ML systems may not inclusively serve all subgroups unless explicitly designed to do so, according to Wynn.
“We have to start at the beginning and work our way through. Part of responsible AI is to make sure that we’re training our developers, product managers, and data scientists to ensure individual bias is not unduly impacting the way we evaluate or design a system,” Wynn said.
The other important focus area is balancing the need for speedy innovation with the necessary guardrails. She’s guided by AWS’s commitment to developing fair and accurate AI and ML services that provide customers with the tools and information needed to build AI and ML applications responsibly.
AWS has built a suite of tools and resources to help customers build AI responsibly, including Guardrails in Amazon Bedrock, which helps AWS customers implement safeguards tailored to their generative AI applications and responsible AI policies; and Amazon Titan Image Generator, which helps customers rapidly produce and enhance high-quality images at scale, all of which contain an invisible watermark to help reduce the spread of disinformation. AWS is among the first model providers to widely release built-in invisible watermarks that are integrated into image outputs and designed to be resistant to alterations.
According to a late 2023 survey conducted by AWS in partnership with the global research company Morning Consult, responsible AI is a familiar topic to business leaders and nearly half of respondents said their companies would invest more into responsible AI in 2024. Wynn said the survey findings also indicate the shared roles we all have in advancing responsible AI and that responsibility doesn’t solely fall on one organization or individual.
“It may seem like a small step, but I often tell customers to invest in internal education around responsible AI, which signals the start of a transformational approach in how their businesses use AI internally and externally,” Wynn said in response to the survey findings. “It’s important to drive a cultural change in which everyone across the organization is aware of how to use AI responsibly in addition to implementing a strategic framework and necessary guardrails to mitigate unintended impact.”
“Roles like mine, in responsible AI, are essential. We need greater diverse representation in this field. We need people that are committed to inclusion, willing to challenge the status quo, and work toward equitable outcomes for all.”
“We have to start at the beginning and work our way through. Part of responsible AI is to make sure that we’re training our developers, product managers, and data scientists to ensure individual bias is not unduly impacting the way we evaluate or design a system.”
Making a difference
For Wynn, there’s an extraordinary amount of personal and professional fulfillment she gets out of her work. Given the scale and scope of AWS, she feels immense pride in impacting how responsible AI makes a positive difference around the world and gives a voice to the voiceless.
"It's rewarding to know the work that I've initiated and led are the right things to usher us into a better future — a future where the way we build is inclusive and responsible," she said.
To drive greater change, Wynn created an AWS Responsible AI Champions program to educate technologists to become ambassadors for responsible practices. Responsible AI Champions is a global team of experts who can help customers look around corners, understand their AI workloads, make informed choices concerning governance and operational processes, and help them put responsible principles into practice.
“It’s a cohort-style program that brought together people across multiple disciplines and roles to learn about key responsible AI considerations, as well as AWS Responsible AI strategy,” Wynn said. “The participants become ambassadors for responsible AI, taking what they learn to conversations on their teams and with their customers.”
Because responsible AI is a fairly new and emergent area, Wynn said it requires constant monitoring. It’s impossible to keep up with it all globally, so she uses technology to stay informed, such as news alerts and generative AI summarization so she can learn the highlights of the day.
“The pace of technological advancement is producing new areas of concern, but not every concern that emerges has answers or solutions,” she said. “So, I spend time in discussions, roundtables, collaborating with other experts across different disciplines to explore approaches and to challenge my thinking. The result is new best practices, recommendations, more areas of research, and often times, more questions that are necessary to help us get to breakthroughs.”
Because the stakes are so high when there’s irresponsible use of AI, Wynn hopes young people thinking about their future career paths will consider this exciting field.
“Roles like mine, in responsible AI, are essential. We need greater diverse representation in this field,” she said. “We need people that are committed to inclusion, willing to challenge the status quo, and work toward equitable outcomes for all.”
Interested in AWS?
We’re always glad to connect with talented people. Tell us a bit about what you want to do and we’ll keep you posted on relevant roles and what we’re building at AWS.
Stories we think you'll like
Article title orem ipsum dolor sit amet, consectetur adipiscing elit
Curabitur congue et est vel scelerisque. Mauris efficitur non metus id maximus. Donec aliquet, libero ac semper imperdiet, tortor eros facilisis velit, sit amet egestas tellus orci non libero.
Read more »
Article title orem ipsum dolor sit amet, consectetur adipiscing elit
Curabitur congue et est vel scelerisque. Mauris efficitur non metus id maximus. Donec aliquet, libero ac semper imperdiet, tortor eros facilisis velit, sit amet egestas tellus orci non libero.
Read more »
Article title orem ipsum dolor sit amet, consectetur adipiscing elit
Curabitur congue et est vel scelerisque. Mauris efficitur non metus id maximus. Donec aliquet, libero ac semper imperdiet, tortor eros facilisis velit, sit amet egestas tellus orci non libero.
Read more »