We Don’t Have to Choose Between Ethical AI and Innovative AI

We keep hearing about how AI is going to steal women’s jobs, proliferate racial bias, make the rich richer and the poor poorer. 

And if we focus solely on that fear, it very well might. 

As the founder of Girls Who Code, I know as well as anyone the risks technology poses to the most vulnerable among us. But I’ve also seen how, when we’re distracted by doomsday, we miss incredible opportunities to help those same communities.

That’s why I believe the next generation of AI will close inequality gaps—if we stop fixating on how it will widen them.

I know it because we’re already seeing it. Take education: After ChatGPT entered the zeitgeist in late 2022, many schools rushed to ban it to prevent cheating. Now, some are walking back that decision. As New York City Public Schools Chancellor David Banks, who oversees the country’s largest school system, put it: "the knee-jerk fear and risk overlooked the potential of generative AI to support students and teachers.”

And that potential is massive. Already, AI tutors are helping students who might otherwise be unable to afford one-on-one support. Tools like Avatar Buddy, which is designed with input from low-income students, are helping them boost their math grades, learn geometry, and understand Shakespeare. Khan Academy is testing an AI chatbot tutor, Khanmigo, that supports student learning with open-ended questions. 

Is there still reason to worry that students could misuse ChatGPT to cheat? Of course. But when students stand to gain so much, the solution is to teach them to use AI responsibly—not back away from it entirely.

The benefits of AI outweigh its drawbacks

In short: the pros outweigh the cons. After COVID-19 disproportionately harmed students of color and lower-income students, widening the achievement gap in education, these AI tutors could very well shrink it.   

AI isn’t just advancing equity and accessibility in education. We’re seeing it do wonders for people with disabilities, helping them live their lives more independently and breaking down barriers to employment. Engineers are using AI to build more resilient infrastructure—which poor communities need more than ever in the face of climate change. We’ve seen incredible advances in healthcare—including AI helping doctors diagnose patients and accelerating drug discovery, which could lower financial barriers to care and save lives. 

And, starting today (Dec. 5), AI is going to support a group of people our country often puts dead last: moms. My organization, Moms First, is rolling out PaidLeave.ai, a tool to help New York parents access their paid leave benefits. We hope, one day, to expand it across the entire country.

Read More: AI Should Complement Humans at Work, Not Replace Them

How exactly does paid leave advance equality? Because there’s no federally guaranteed paid family leave in the U.S., many workers—especially low-income women—take unpaid time off to care for family members or themselves. That time comes at a price: lack of paid family and medical leave costs working families $22.5 billion in wages every single year. 

Even for people lucky enough to live in one of the 13 states or in Washington, DC, that has paid leave benefits, there are still sizable obstacles to accessing care. So even if parents know about their benefits, they still likely have to navigate dense, government-penned legal jargon. Remember, these are parents who barely have a moment for themselves, let alone hours to comb the internet. The reality is, many people simply give up.

This is exactly the kind of issue AI can help solve. PaidLeave.ai can streamline a dense tome of government paperwork into a simple, user-friendly experience. Parents can ask as many questions as they want, in many different languages, as if they were texting a friend—and PaidLeave.ai can come back with answers.

AI can help the public sector deliver services

And it’s just the beginning. Finally, it seems the public sector is waking up to the many ways AI can and will impact their work. Paid Leave.ai builds on the success of New York City’s MyCity Business AI chatbot, which helps people start and grow small businesses by delivering trusted information from more than 2,000 web pages.

We’re starting to see investments like these across the country. A recent report found that around 70% of government leaders have created teams to develop AI policies and resources. And federal agencies are working on roughly 700 uses cases for AI tools, with particular enthusiasm from the Department of Energy and Department of Health and Human Services. It’s early days, but it’s an encouraging sign of public-sector-spurred progress to come. 

Of course, AI is not without risks. While we’ve helped make our chatbot more reliable by training it on government sources, including New York’s Paid Family Leave website, we’ve also heard stories of chatbots spitting out information that simply isn’t true. That’s why, as we continue to develop these technologies, we must make sure they can still “think” creatively while knowing fact from fiction—and why it’s so important we test them and regulate them in tandem, sooner rather than later. 

If we are going to build trust in AI among people who have never used these tools before, then the people these tools are built to support must be at the table, helping design AI solutions from the start. In our case, that means centering the perspectives of moms—specifically single moms, moms of color, and moms who are hourly workers. 

But this applies everywhere. If we are intentional about harnessing the power of AI to empower the most vulnerable, we lift up everyone. That’s what the great minds of tech should be focused on right now.

To ensure anyone can create these tools, we must give everyone access to them. So let’s learn from the mistakes of technologies past—and the digital divide that emerged from them—and get AI in the hands of women, young people, people of color, and low-income communities while the technology is still in its relatively early stages.

And let’s be clear: we don’t have to choose between ethical AI and innovative AI. We can develop new use cases to lessen inequality—and, at the same time, train employees and students on AI ethics. 

Right now, we have an opportunity to solve some of the world’s biggest problems and help the world’s most vulnerable communities. But it will require a willingness to take bold swings, put people first, and cooperate across sectors, industries, and political parties. 

Because at the end of the day, our AI is only as good as we are.