KOSA AI

Artificial intelligence (AI) constitutes an essential part of today’s economy, but is riddled with biases that disadvantage BIPOCs, women, and other minorities unfairly. As a result, many cannot access the credit they need, do not receive healthcare they qualify for or get rejected from jobs unfairly. At KOSA AI, we have developed a proprietary algorithmic auditing solution to help businesses identify and minimise the biases inherent in their AI-powered decision-making models. Our solution is sector-agnostic and has multiple use cases including fraud detection, insurance, healthcare delivery, credit risk assessment, customer lifetime value, and talent acquisition. By identifying and mitigating biases in AI and ML processes across industries, our solution will reduce the hidden inequalities that today affect millions of people and enable everyone to access the services and jobs they want irrespective of race, gender, age or other status. 

AI constitutes an essential part of today’s economy and powers thousands of decision-making tools. However, AI is riddled with biases that disproportionately impact women and BIPOCs in applications as varied as healthcare, financial services, policing, and hiring. Facial recognition technology detects male faces with a 99% accuracy rate compared to 65% for black women; similarly, the algorithm behind the Apple Card was shown to give 20x more credit to men than to women. Last year, a healthcare algorithm sold by Optum was found to show drastic racial bias, denying care to 50% of qualifying black patients. 

Numerous factors cause bias in AI. These may occur due to historic prejudices that are reflected in the training datasets and inputted into the model (built-in bias); when a credit provider uses employment history and prior access to credit as factors to determine creditworthiness, it automatically imports racial biases into its credit risk assessment tool. Another factor is related to the wider configuration issues associated with the model that are then amplified as the algorithms evolve (AI-induced bias). 

AI biases could be mitigated with a reliable detection solution, but today only open-source library tools and expensive consulting services are available.

KOSA AI’s algorithmic auditing software enables large enterprises in regulated sectors, like healthcare, financial services, and government, to understand and mitigate bias risks within their AI systems. We offer customers a fully-automated solution that seamlessly integrates into a company’s existing AI infrastructure. Our multi-stakeholder product comprises of 4 steps that provide support across the whole ML lifecycle: (1) it assesses and mitigates the biases in current ML processes; (2) it audits the model to assess the human impact and automate compliance checks (including adherence to the Digital Service Act) before deployment; (3) it explains why the model behaves the way it does; and (4) it builds a model monitor which tracks drifts or malfunctions and allows developers to fix vulnerabilities. 

We realise that building responsible AI is a team effort; this is why we have developed tools for all stakeholders, from executive team to product managers and developers. The software outputs an evaluation for the technical development team to identify and understand the bias within their systems and a quantifiable financial assessment targeted at non-technical stakeholders to grasp the missed opportunity that the bias brings to their business.

KOSA AI’s mission is to make technology inclusive of all people. We work to support SDG 10 to reduce inequality, in particular SDG 10. By 2030, empower and promote the social, economic and political inclusion of all, irrespective of age, sex, disability, race, ethnicity, origin, religion or economic or other status. As seen above, the wide use of AI decision-making technology in financial services, healthcare and public sector fuels inequality for various minority groups. Because of biases inherent in AI-powered decision-making tools, BIPOC are denied access to quality healthcare, insurance and financial services, fair justice and policing, job opportunities, and more. As a result, the population we target are all those individuals who are negatively impacted by biases present in AI and ML, especially BIPOC.

Due to the nature of our solution, we work with large enterprises and not our target beneficiaries. However, we do everything we can to ensure that our software is having the impact we aim to achieve. For instance, we conducted a proof of concept with healthcare applications which was shown to increase BIPOC healthcare service utilisation by 30%. In addition, we work with a number of research institutions to increase knowledge and awareness of AI bias mitigation and ethical bias; we want to make sure that AI-driven inequalities are better understood to develop solutions that respond to the needs of those impacted. 

Finally, we leverage our distributed model to improve our software and increase our impact. Through our presence in Africa we can (1) access diverse training data sets which we can share with companies outside of the continent who require more representative datasets; and (2) build a data bank with more diverse data sets; and (3) increase the diversity of our team by mobilising African talent.

The Challenge seeks solutions that will help end the disparities between BIPOC and white communities in the US. It requests solutions that are technology- or innovation-based and use data science for positive change to support and liberate BIPOC communities. Our solution is a technology-based innovation that enables companies to identify and correct biases in their AI and ML that prevent BIPOC from accessing the services, products, and jobs they need. Through our technology, credit agencies can build fair credit profiles and healthcare providers can give adequate healthcare, all while improving their hiring practices to include and support BIPOC. 

We launched a minimum viable product (MVP) earlier this year and are currently piloting our solution with 5 partners and customers across healthcare, technology, services, government, and education. We expect to convert at least 2 of our 5 piloting organisations into paying customers and plan to fully launch our product by the end of 2021. We currently have 5 employees at KOSA AI, including the Co-Founders, and foresee growing to include a team of enterprise sales specialists to help us expand our client base in our next step of development.

KOSA AI is the world's first SaaS solution to AI bias auditing. Today, most companies and organisations that use AI and ML in decision-making tools don’t conduct algorithmic auditing and monitoring for potential biases primarily due to two barriers: (1) the misconception that bias auditing is an expensive cost and (2) the lack of buy-in from non-technical stakeholders. AI managers acknowledge that fair AI is a priority for them, but are often held back because they lack the right tools and fail to gain leadership engagement. 

Our solution is innovative because it addresses precisely these two inhibitors. It is half as expensive and 10x more efficient than current alternatives on the market. It targets every AI and ML stakeholder across the organisation with a set of relevant tools and dashboards that are didactic and easy to use, increasing understanding of biases in data and algorithms and detailing the financial benefits of correcting the biases. In addition, by leveraging our presence in Kenya, we are in a position to access diverse training data sets which we share with companies outside of the continent who require more representative datasets. 

Ultimately, we enable businesses to improve the quality of data that is fed into their decision-making tools to make them fairer while increasing their revenue stream; thanks to us, companies can be more profitable by being more equitable. We expect to mainstream AI data fairness across organisations and sectors, catalysing broader positive impact. 

As previously mentioned in the application, we work with organisations that offer services to our end beneficiaries. At this stage of development we are therefore actively tracking and forecasting our current customers while estimating the number of beneficiaries impacted. 

Current customers: 5 customers, serving roughly 0.5 million people

Customers in 1 year (2022): 20 customers, serving roughly 2 million people

Customers in 5 years (2026): 600 customers, serving roughly 50 million people

Currently, we are tracking the following: 

• Number of customers (active vs non-active) per industry vertical i.e. Healthcare and Life-sciences, Banking and Financial Services and Government and Defence.  

• Email tracking: percentage of customer gain from email reach (cold and warm introduction)

• Number of biases detected in enterprise systems, in percentage terms per use case

• Monthly burn rate which is currently only USD 5K. Being a software solution, we are lean by definition with a low OPEX. 

• 20 monthly stakeholder interviews with non-technical users e.g. Head of AI, Product Managers and technical users e.g. data scientists and engineers. Feedback from user interviews will help shape product offering and value proposition. 

• Product feature usage

In the long run, we would like to measure the following: 

• Number of people indirectly impacted (we are currently in the process of determining a viable process to do so)

• Monthly active sales per use case/per industry vertical

• Other sales and marketing KPI's

• Economic activity e.g. liquidity ratios

• Value determined from key partnerships e.g. percentage of useful research that can be (1) incorporated into the product solution and (2) impact driven on target population i.e. minority groups

4 full time

1 part time

KOSA AI was founded by entrepreneurs passionate about our mission to reduce inequalities exacerbated by technology.  With multidisciplinary skills across software development, data science, business, and healthcare, our small team has the technical and commercial expertise to deliver our solution. Key profiles below:

 Co-Founder/CEO Layla Li is a Boston University and Harvard University graduate with 7+ years experience building technology solutions at companies including Tesla and Philip. Layla is passionate about AI bias because of her own experience as an impacted minority group; she built a strong expertise in AI bias during her time developing automated decision-making systems for multiple international organisations. 

Sonali Sanghrajka, Co-Founder/Chief Commercial Officer, has 10+ years’ experience in the healthcare sector, driving brand and commercial strategies for products worth $500 million. Sonali is passionate about AI bias as she has first handedly worked with patients who have been directly affected by bias through care delivery. And through her consulting services, she has been privy to the challenges that AI companies have in penetrating the African continent with their AI product/solution offerings. 

 A leader in the AI space, Chief Product Officer Kunal Sawarkar is a Harvard graduate in Applied Statistics Kunal with 15+ years’ experience in data science and product development at IBM, ADP, and Aon.

At KOSA, we embrace diversity -- in fact, diversity, equity, and inclusion are both essential values and foundational pillars of our mission and strategy. Our Co-Founders are women from diverse backgrounds, Layla is Chinese, Sonali is Indian-Kenyan, while our CPO is Indian -- and believe that this diversity is essential for us to achieve our mission of reducing technology-fuelled inequalities. In addition, our presence in Africa enables us to actively recruit employees from the continent which further increases the diversity of our team.

We believe Solve’s mission to reduce inequalities between BIPOC and white communities in the US directly aligns with our own mission. Therefore, we seek help from Solve to support the following:

Scaling and partnering: As we launch our MVP in Europe, we are looking for organisations to support us and partner with us to pilot our product across the US. We are looking for companies across Healthcare and Lifesciences, BFSI, and Government and Defence. We would also welcome the opportunity to work with academic and research institutions to further increase the reach of our product.  

Grant funding: We are looking for in-kind contributions that will subsidise the cost of piloting and testing our product at a discounted price across organisations in the US, with the objective of converting these partners into paying customers. We would also welcome grants to enable collaborations with academic institutions and fund research projects that advance the field of ethical AI and bias impact. 

Impact measurement: We are seeking support to develop an impact measurement framework to better understand the needs of our end beneficiaries. We have focused on the development of our product to ensure maximum usability and results for our customers; however, we would like to develop a system that enables us to better study our target beneficiaries and integrate their needs into our development process. 

Networking and mentorship: We welcome the opportunity to network with institutions and individuals that share the same vision of reducing bias and creating more responsible and trustworthy AI.

Public Relations: As a young company, we look to increase our marketing efforts and visibility across US media to support our partner-acquisition objectives.

Monitoring & Evaluation: As mentioned above, we are looking to strengthen our practices around impact measurement.

Product / Service Distribution: Again as mentioned above, we welcome opportunities to expand our partner and customer base across the US.

We welcome the opportunity to partner with a number of organisations in the MIT and Solve community, including:

Bill and Melinda Gates Foundation: With its investment in the Wadhwani Institute for Artificial Intelligence Foundation, the Gates Foundation has started to look into explainable and responsible AI and its applicability in global health contexts. A partnership with the Foundation would enable us to deploy our solution across a number of projects and settings improving care delivery for thousands of individuals.  

Novartis Foundation: The Novartis Foundation works with governments around the world to shape the policy needed for safe, effective, and inclusive integration of AI into health and care. As we look to lead the conversation on responsible and inclusive AI standards, we see a number of ways in which we could collaborate to ensure governments are taking the right steps to integrate inclusive AI in healthcare settings.

Oxford Sciences Innovation: We see Oxford Sciences Innovation and its portfolio companies as (1) potential research partners to advance research in ethical AI and bias detection tools, and (2) potential implementation partners to pilot our solution across a wide range of healthcare applications. 

The Robert Wood Johnson Foundation Prize seeks solutions that address the health and well-being of people in the US, with a special focus on health equity. Our solution seeks to improve health equity for BIPOC across the US by providing AI bias detection and mitigation strategy to healthcare companies using AI-powered decision-making tools. The healthcare industry sees a large number of applications of AI solutions across provider settings, including care delivery (e.g. Arterys), medical imaging and point of care diagnostics (e.g. Butterfly Network), and clinical decision support systems (e.g. Babylon Health). However, AI bias is pervasive across enterprise data models; this impacts BIPOC’s ability to access quality health services. As seen above, UnitedHealth’s Optum algorithm, serving 200 million people, showed drastic racial bias, denying care to 50% qualifying black patients. At KOSA AI, we seek to reduce these inequalities by equipping businesses with our AI bias detection and mitigation technology. With support from this prize, we can accelerate the deployment of our technology in healthcare institutions across the US and start improving healthcare delivery to those who so far have had this essential service repeatedly denied.

The HP Prize for Advancing Digital Equity seeks solutions that advance inclusion in communities across the US and globally. Our solution advances digital equity by increasing inclusion of traditionally underserved and excluded individuals by identifying and mitigating the biases inherent in AI-powered enterprise decision-making tools. Biased data sets and algorithms have continuously denied access to quality jobs, credit lines, healthcare services, and more to individuals who would otherwise qualify for them. By using our algorithmic auditing software, companies and organisations can now identify and mitigate these biases to ensure that all qualified individuals, irrespective of race, gender, age or other status, can access better jobs, services, products, and governing systems. With support from this prize, we can accelerate the deployment of our technology across the BFSI and recruitment services sectors to improve access to financial services and insurance and strengthen economic opportunities for thousands of communities across the US and globally.

The Innovation for Women Prize seeks solutions that use innovative technology to improve quality of life for women and girls. We have developed a novel software solution aimed at bettering job opportunities and products and services available to women by identifying and mitigating gender biases in AI-powered decision-making tools. Current facial recognition technology detects male faces with a 99% accuracy rate compared to 65% for black women; similarly, the algorithm behind the Apple Card gives 20x more credit to men than to women. These are only two examples of how AI biases disproportionately affect women in ways that are often invisible, but hundreds of other sample cases, from salary to healthcare delivery and safety, exist. By equipping organisations with AI bias detection and mitigation technology, we enable them to develop more inclusive product offerings and recruitment practices and ultimately improve the quality of life for millions of women and girls.  With support from this prize, we can accelerate the deployment of our technology across companies and organisations globally and ensure that millions of women have access to better jobs and quality financial and health products.

The AI for Humanity Prize is looking for solutions that leverage data science, AI, and/or ML to benefit humanity. At KOSA, we have developed an innovative solution that seeks to de-bias the AI- and ML-powered decision-making tools that thousands of companies use to develop and deliver products and services to consumers as well as hire employees amongst other functions. Our technology can also be leveraged to improve AI and ML use in judicial systems and policing. KOSA's automated responsible AI system (ARAIS) helps companies build technology more inclusive of all ages, genders, and races. Therefore, our algorithmic auditing software not only leverages advancements in data science and AI to benefit humanity, it also seeks to enable large companies and government agencies to do the same. Although so far we have partnered with organisations in BFSI, Healthcare and Life-sciences, Government and Defence, and Education, our solution is sector-agnostic and has a number of use cases, including fraud detection, insurance allocation, healthcare deliveries, credit risk assessment, talent acquisition, and customer lifetime value. With support from this prize, we can accelerate the deployment of our technology across companies and organisations globally, while increasing research-focused projects that seek to advance the fields of ethical AI and AI bias mitigation. This will enable millions of people to access a better quality of life and help create a thriving and equitable future for all.