X
Finance

Blockchain aims to solve AI ethics and bias issues

Many Machine Learning and AI algorithms are centralized, with no transparency in the process. Now a blockchain-based start-up aims to improve transparency bias in business workflows
Written by Eileen Brown, Contributor

The global AI market is projected to reach $266 billion by 2027. According to IDC, worldwide spending on AI is expected to double in four years, reaching $110 billion in 2024.

However, the rapid increase in the global adoption of AI has led to concerns over privacy, fairness, and transparency.

Gothenburg, Sweden-based tech start up Unbiased aims to solve these issues. It is building solutions to solve ethical challenges in AI and big data using blockchain technology.

The company has announced the launch of its Data Marketplace tool. The tool is intended to be a 'one-stop-shop B2B solution'  for enterprises working with AI and Machine Learning applications.

The solution includes privacy-centric and decentralized AI development tools to companies who are working with AI and Machine Learning applications including data collection, annotation, labelling and analytics which are all recorded on the blockchain.

The project was introduced in beta in March 2020, and is now live for commercial use.

Currently, most dataset generation tools for training supervised Machine Learning and AI algorithms tend to be centralized, with no transparency in the process. This can bring biased results as the algorithm learns, and can be kicked off track by rogue data.

The start-up uses the Telos blockchain to track all actions on the Data Marketplace platform, including the creation of projects and tasks, worker contributions, and exchange of data. Unbiased chose Telos' governance engine, Telos Decide, for integration into business workflows.

The Unbiased Data Marketplace is supported by Unbiased WorkForce, a crowdsourcing platform with data collection and annotation tools needed to source AI and ML datasets. The platform currently supports over 20 project and task types.

 You can contribute and earn money for your work by signing up for Unbiased through the App Store, Google Play, or the web app, finish the qualifier tasks, and start to work on paid tasks.

The first sets of tasks are live now with initial projects focusing on data collection and audio transcription.

Sukesh Tedla, Founder and CEO of Unbiased said: "Unbiased uses the Telos blockchain to improve transparency and integrity, allowing users to trust and validate whether their dataset annotations are happening according to set requirements."

Gartner research has revealed that organizations deploying AI grew from 4 to 14 percent over the past two years so protecting data transparency, trustworthiness, and ethics in the multi-billion dollar AI industry is critical.

The European Union are researching the far-reaching impacts of AI and how to best address transparency challenges.

Perhaps the immutable ledger and transparency of the public blockchain will provide the decentralised, trusted solution AI and ML desperately needs.

Editorial standards