A group of data scientists and AI specialists are working to develop an AI checklist. The proposed guideline is to ensure the safe development of AI-centered products.
The World Ethical Data Foundation (WEDF) has come up with a framework of 84 questions. It's developed by the group's members, which stand at 25,000 strong. The WEDF has specialists working in tech giants like Google, Meta, and Samsung.
The WEDF’s questions would act as a guide for developers. It was already released in an open letter form. It entreated AI developers to consider the questions before starting any project. The group believes the checklist would help ensure safe systems are always maintained.
The WEDF stated its belief that artificial intelligence can become a force for good. But this will only happen if it’s developed and used in an ethical and responsible manner. The group also recognizes the extent AI systems can impact communities. Both in positive and negative ways. They also acknowledge that these systems can become subject to abuse and misuse. These can lead to unforeseen consequences. That’s why the WEDF came up with a proposed framework for AI experts to follow.
The suggested AI guidelines are all founded on four key principles. These are respect for human dignity and autonomy. The principles of fairness and justice, and transparency and accountability are also considered. So is the principle of safety and security.
This checklist of sorts covers different elements of the AI lifecycle. For example, data collection, processing, and analysis. It also considers aspects like modeling, testing, and deployment.
Here are some of the questions included in the proposed framework:
Am I pressured or rushed to input data from dubious sources?
What is the model’s intended use once it’s trained?
How will I make sure the model doesn’t discriminate against any group or individual?
How will I track the performance and impact of the model?
How will I retire or update the model if it becomes harmful or obsolete?
There are also questions on the types of data used to train the model. Consideration was also given to data protection laws from around the world.
The WEDF open letter has hundreds of signatures from specialists in the field. The group has also asked the public to send in their own questions. These will be discussed at the group’s next conference.
The AI framework is the brainchild of Vince Lynch. The WEDF advisor described the AI landscape as akin to the Wild West. It’s free and everything is running rampant. But the cracks in the foundation are also starting to show. The framework will help ensure AI development remains ethical.