Objectives for the future of AI
Last updated
Last updated
The number of AI incidents and controversies reported in the AI, Algorithmic, and Automation Incidents and Controversies (AIAAIC) repository has reached an , being 26 times larger than in 2012. Perpetuating stereotypes or discriminating against individuals are just some examples of the profoundly negative impacts that improperly trained AI can have on our society.
With AI , the ethics involved in developing the AI of the future is becoming more and more a societal problem and is no longer a mere academic debate.
In line with the and programmes, Timeworx.io is on a quest to promote Hugging Face’s “AI for the masses!” mantra to the general public. Our vision is to create a space where everyone can contribute to a more ethical, open and mutually equitable future of AI development guided by these three objectives:
AI that is Fair: ML models that are trained on non-diverse and biassed data lead to allocative and representational harms which, in turn, make predictions that disadvantage people and tarnish the image of AI. At Timeworx.io, the approach to data labelling is set to be distinctly different, ethical and innovative. The focus is on creating a decentralised protocol based on openness, diversity, transparency, accountability and mutual equitability.
AI that is Privacy-Enhancing: Data, especially when sourced from users or sensitive sectors, needs to be treated with utmost confidentiality. Ensuring data privacy during the labelling process becomes an imperative. With its foundation on the blockchain, data integrity and security are paramount in Timeworx.io. Furthermore, we are taking one step further with the integration of federated learning. In a completely decentralised manner, data never leaves the user’s phone, rather the ML model is downloaded, trained, and uploaded without leaving any trace of the data it was trained on.
AI that is Trusted: The AI landscape is dominated by large tech companies which drive innovation and dictate the tone. Without any formal training in the ML field, the majority of society is looking up at these key players on the market with awe and appreciation based solely on their influence and visibility. Timeworx.io takes a different, trustless, approach in which no AI (or human for that matter) is considered to be trusted. Our innovative decentralised protocol ensures that the ground truth in data processing is determined through Consensus, and that performance is accounted for using on-chain proofs.
