Winners for Stage 1 of Perlin’s AI 4 All Global Hackathon
In Partnership with the Indian Government & PT Telkom
Working with the Indian Government (NITI Aayog) and PT Telkom (Indonesia’s largest telco with 180m users), we launched the AI 4 All Global Hackathon in November last year. The challenge was set for teams to come up with ideas on:
“How can AI be used to make significant positive social and economic impacts to advance India?”
Starting with India, it is envisaged that pilot programs could eventually be rolled out in Indonesia and other developing economies around the world. Since then, we’ve received a large number of submissions from applicants all around the world. Over 20% of the world’s population is situated in India and Indonesia, and is therefore a compelling case for AI developers and companies to create a positive and meaningful real-world impact.
We’d first like to thank all participants who sent in a submission! We spent the last few weeks assessing all the submissions, and we were delighted to see so many interesting and inspired submissions.
After an exhaustive review of the many impressive Hackathon submissions that came in, we’re now pleased to announce the prize winners for Stage 1:
1st Place: Privacy Research Foundation Team ($8,000 USD in PERLs)
Proposal: Differential Privacy for AI/ML for Healthcare and Government
To really advance the data revolution, secure methods that are mathematically proven must be devised to preserve the privacy of datasets whilst running models on the data. The Privacy Research Foundation Team proposes to build a set of flexible, computationally efficient, and provable privacy preserving tools to enable private/public entities to adopt data practices that preserve user data in a socially responsible and democratic way. This will shift how companies and governments access, analyse and share sensitive data to drive innovation across varying industries.
This will potentially impact:
Individuals: The privacy of personal data is not only an issue in India, but also one on a global scale. The ability to privatise user data without disclosing the actual data is powerful. Users will have the confidence that their data can be used to help facilitate research with minimal risk.
Enterprises & Governments: Restrictions on sensitive data sources, unused or underleveraged can be relaxed or removed — enabling powerful new data-based advancements across every conceivable industry and community (e.g. healthcare, user activity) and research could be made readily more available on complete datasets as opposed to filtered down datasets as a result of potential privacy breaches.
2. Immutable Works Team ($3,500 USD in PERLs)
Proposal: Data Privacy for Training/Running AI models on decentralized networks
To address the problem of data privacy when training and running AI models on a decentralised network, the Immutable Works Team propose a simple, efficient, and secure multiparty computation that is well suited to the use case of AI in decentralised compute networks. Enterprises are required to host the data of their users, which can lead to data leaks and trust concerns from users. Using secure multiparty computations on a decentralised compute network such as Perlin can resolve these issues. Researchers and companies will be able to train, improve, and run AI models without requiring direct access to user data.
This will potentially impact 2 key groups:
Consumers: an estimated 100M Alexa-enabled devices, 375M monthly Siri users, and 141M monthly users of Cortana. Even if a small market of consumer AI is transitioned to decentralised networks in the name of privacy, millions of consumers in India and globally will be impacted.
Researchers: an estimated >300,000 AI engineers in India and globally.
3. Cybernetica Team ($1,000 USD in PERLs)
Proposal: KYC Using Privacy Preserving AI
Using privacy-preserving AI to implement Know-Your-Customer (KYC) that doesn’t create persistent links between identifiers (i.e. potentially breach privacy). The proposed method will not compromise privacy but will implement a KYC AI that combines multiple data sources to create a trust score for an identity (using secure computation techniques).
This will potentially impact every person involved in any payments or telecommunication streams where transactions need to be checked for fraud. That would include the majority of the adult population in India and globally.
Each of these Stage 1 prize winners will be invited to participate in Stage 2, alongside a number of the other most promising Stage 1 applicants. Stage 2 will require applicants to further improve their ideas, develop a demo, and present their submissions to the Hackathon Judges.
Based on the strong field of applicants for Stage 1, we expect to see many exciting and innovative techniques leveraging AI, distributed ledger technology, and privacy preserving techniques. The Stage 2 prize winners will also have a chance to collaborate with Perlin, NITI Aayog and PT Telkom to implement their ideas on a truly massive scale in both India and Indonesia, with a view to global implementation.
Private invitations for Stage 2 submissions will go out on 4 February. The deadline for Stage 2 Hack Submissions is 27 Feb 2019.
Stage 2 winners will be selected by the Hackathon Judges and announced by 15 Mar 2019.