Our 2021 Tech for Good Awards Winners Revealed
The Stack’s 2021 Tech for Good Awards drew some inspiring applications.
While innovative uses of technology in the third-sector are always positive to see, what excited us was the extent to which genuinely transformational technologies with a social or environmental purpose are increasingly likely to have a life of their own as business propositions — amid a world in which environmental, social and governance (ESG) requirements are top of mind for both policy makers and a growing number of large investors.
So we’re delighted after some tough judging to announce the three winners of our inaugural, 2021 Tech for Good awards, generously sponsored by Red Hat and Rainmaker Solutions at launch earlier this year.
Here’s our first winner
Winner: Fathom (SME)
Fathom, born out of the hydrology department at the University of Bristol, is extending comprehensive flood risk data to traditionally data-scarce regions. The growing need for this barely needs explaining: most European readers will be aware of the floods that killed hundreds across Germany and Belgium this summer.
(Prior to that incident the Ahr river’s previous record high-water-level had been 3.71 metres: judged to be a 1-in-100-year event. In July 2021, it peaked at an estimated 7 metres – “estimated” because the measuring gauge was torn away during the surge. As just one example of the need for improved modelling globally, it’s a visceral and recent one. Globally in 2020 meanwhile, five of the six deadliest events were seasonal floods.)
To model flood risk accurately at the scale Fathom is achieving is an enormous endeavour, requiring a suite of technological tools which are the culmination of decades of collaborative work by scientists all over the world. The key piece is globally-seamless terrain data, collected by satellites and painstakingly corrected by the scientific community. Marrying data availability, rapid algorithms, and greater availability of computer power, Fathom has helped ensure that no patch of land globally is without a flood model and through the public release of Fathom’s US flood data at www.floodfactor.com, we can already begin to see how democratising availability of the crucial information is allowing individuals and organisations to build resilience to this growing problem.
Congratulations to the team at Fathom!
Winner: Persefoni (medium-sized organisation)
Persefoni has built a fully auditable carbon footprint management platform, that includes planning and forecasting tools, which allow enterprises to run carbon footprint scenarios and manage their decarbonisation efforts through ongoing forecasts; empowering emissions reductions programmes.
By treating carbon inventories and transactions similar to how financial ERP systems manage financial accounting and transactions, the platform lets users quickly and easily report their data into multiple frameworks and standards, including SASB, TCFD, SECR, GRI, and CDP. Persefoni is also the first platform to codify the PCAF Standard for calculating measure financed emissions. As pressure from regulators and institutional investors grows on enterprises to more clearly report the environmental impact of their operations, platforms like Persefoni’s are going to be a vital enabler of the kind of transparency that will empower stakeholders to understand and ultimately reduce the environmental impact of their operations.
Congratulations to the team at Persefoni!
Winner: Feedzai — large enterprise
Feedzai’s Fairband brings practical methodologies and tools for real-world practitioners into the world of algorithmic fairness. With racial and gender discrimination in recidivism assessment, job-applicant screening tools, access to credit, or medical diagnosis all widespread (among many other examples), the importance of tackling AI bias is a colossal one; deployable tools to help tackle it are less ubiquitous.
Fairband, an algorithm that discovers much less biased machine learning models automatically and with zero additional model training cost, aims to help tackle this gap — and has been shown on average, to increase model fairness by 93% with only approximately a 5% drop in model accuracy.
Ingeniously, it can be used with any fairness metric (e.g. equal opportunity or false positive rate), any model metric rate (e.g. precision or recall), any sensitive attribute (e.g., age, gender, race), any algorithm (e.g., neural networks, random forests), either model settings (e.g., punitive such as recidivism forecasting or assistive such as grant approvals) and has almost universal domain applicability. Within a financial context, it can be used to help ensure access to financial services is not being disproportionately held back for people due to race, age, residence, profession, or employment status.
“By incorporating Feedzai Fairband in our product and allowing other types of organisations to use it via open-source — which the company is currently working on — we are potentially impacting over 800 million of our customers that will face less discriminatory access to financial services” noted Feedzai in its application. For example, in a client (a retail bank) where Feedzai tested Fairband, close to 1000 people per month that would be unfairly denied applications were now recommended for account opening approval.
(Data scientists and other AI specialists interested in looking under the hood can do so via this research paper). It’s a robustly researched project from Feedzai that we would like to see built on and widely deployed — expect to hear more.
Congratulations to the Feedzai team!
The Stack’s Tech for Good awards will be back, bigger, bolder, and with more categories in early 2022.