As artificial intelligence (AI) becomes increasingly interwoven into the fabric of our lives—from Alexa providing a daily weather forecast to applying machine learning to better diagnose and treat diseases—there is a growing appetite to harness its power for good.
At Accenture, we have embraced the potential of AI and other emerging technologies, including extended reality, blockchain and IoT, to address societal challenges. Over the past 12 months, we have brought together Accenture Leaders in Innovation including Accenture Labs and emerging technology and Corporate Citizenship with our Skills to Succeed nonprofit partners, jobseekers and entrepreneurs to push the boundaries of AI by applying it to:
Improve access to upskilling, employment and entrepreneurship opportunities
Help vulnerable groups (e.g., opportunity youth, refugees and migrants, persons with disabilities, incumbent workers at risk of displacement) thrive in the digital economy
What have we learned so far about how to maximize the upside and minimize the downside of AI when applied for social good?
Fit for purpose. When designing an AI solution, start with the end in mind. Many innovation projects lead with technology rather than focusing on the specific challenge that must be addressed. This is like a hammer looking for a nail.
Recently, a group of high school students told one of our U.S. Skills to Succeed nonprofit partners that it was difficult to locate program-specific information on their website. After evaluating ways to fill these information gaps, we determined that an AI-enabled virtual agent was the best fit to provide personalized answers at key points in the early interest, application, and enrollment process.
People at the center. Involving the end-user from the very start of the design process is the best way to ensure that the solution addresses a real need and has a meaningful impact.
Human-centered design is integral to our collaboration with nonprofit Leonard Cheshire Disability (LCD), one of the United Kingdom oldest and largest disability-focused organizations. After learning that low-vision jobseekers found the registration stage of their journey particularly challenging, we ideated on how to improve this process.
Together with LCD we developed an AI-enhanced way to sign up for livelihood support service and skills training, creating a ChatBot, Nandini, to guide a visually impaired person through the registration process. Our intent: "With guidance from a virtual assistant, persons with disabilities can access and easily complete the registration process by themselves, so that they can feel confident and happy about taking the first step towards getting a desired job."
As another example, we followed a youth-centered approach when developing the Accenture Intelligent Space Exploration tutorial, in support of our partnership with Hour of Code. This first of its kind, AI-enabled tutorial provides a youth-friendly, gamified approach to learning about AI.
To help ensure the tutorial’s effectiveness, we worked with students across a range of ages, nationalities and backgrounds to design, prototype and test our solution. We are proud to say it is “kid tested and approved,” and has been played on code.org by more than 70,000 people.
Make it stick. Arguably the most time-consuming, challenging and costly part of taking an idea from concept to reality is building the ecosystem that will make it sustainable at scale. We recently developed a prototype for an AI-enabled chatbot, TESSA, with Plan International in the Philippines, that uses Facebook Messenger to support youth during and following employment training. The chatbot helps users identify skills, build resumes and identify training and employment opportunities. Leveraging a social media platform that a high percentage of young people in the Philippines already use will make it easier to scale.
Additionally, Accenture Labs in India collaborated with the National Association for the Blind to develop Drishti, an AI-powered solution for the visually impaired. We aim to expand Drishti’s reach and long-term impact by facilitating the adoption by other organizations.
"Raise" AI with responsible use of data. Combining human and machine intelligence to create machine learning models establishes positive feedback loops that can improve results over time. Take for example, skysthelimit.org, which Accenture built with Youth Business USA to connect young entrepreneurs from under-represented communities with experienced business advisors and mentors. To facilitate these connections and help ensure a “good fit” for each entrepreneur and mentor pairing, we are developing an AI-enabled matching system that can learn and improve how matches are made.
As the algorithm learns, we are careful to prioritize data quality, veracity and privacy. Since algorithms "learn" patterns from data, it’s essential to select data sets that do not reinforce existing biases and to design algorithms that are easily understood and transparent about their underlying decision models.
Consider how an AI-enabled hiring platform that masks an applicant’s gender and ethnicity could help level the playing field for traditionally underrepresented candidates. By combining human ingenuity with groundbreaking technologies, we are helping to minimize the conscious and unconscious biases that influence candidate screening.
With each new prototype, we design a "responsible use plan" upfront, think through any legal, ethical or employment impacts of the technology and revisit the plan often throughout the lifecycle.
Test, test and test again. Testing solutions early and often is critical. The goal is to follow an iterative process where a solution evolves several times.
For example, Accenture’s Emotions Analytics tool is an AI-powered, 24/7 interview practice aid that uses emotional intelligence, voice and facial recognition software to prepare jobseekers for the interview process.
When we first tested the tool with young people in the United Kingdom, we found that the feedback was often too raw for individuals who had never been through interviews or received on-the-job feedback. In our second iteration, we incorporated a coach to help prepare users, and saw positive results when we rolled out the modified solution with women reentering the workforce in India.
Further, Accenture partnered with Age UK to develop HomeCare, an AI-enabled companion that provides the elderly with support and assistance in areas such as scheduling health appointments, medicine reminders, grocery shopping, exercise and staying connected with loved ones. Accenture conducted a three-month pilot program to identify opportunities to improve the platform. Overall, participants had a positive response to HomeCare, including reinforced autonomy.
Innovate with Us
We are committed to applying these principles to ongoing social innovation and to sharing what we learn. We hope this will inspire others to seek new ways to address societal challenges using emerging technologies … and to share their insights and results along the way.