AI Governance for Supportive and Public Systems: How Policy Makers Can Strengthen Essential Digital Competencies
In the ever-evolving world of technology, a significant focus is being placed on the development of assistive AI systems. These systems are making a tangible impact in various assistive technologies, such as prosthetic arms, smart wheelchairs, and robotic kitchen assistants.
The need for these technologies is immense, with approximately 285 million people worldwide living with visual impairments and around 1% of the global population being autistic. However, there is a systemic data gap across text, visual, and 3D spatial data for assistive AI, with few models trained on data from people with disabilities.
To address this issue, governments are prioritising investment in energy-efficient, offline-capable models optimised for low-power environments through compression techniques, fine-tuning for specific assistive applications, and specialised architectures. This approach aims to make these technologies more accessible, especially in hospitals, rural clinics, or home care settings where power and internet access may be limited.
The safety and security of these assistive deployments are also of paramount importance. Given that they operate in high-stakes environments like hospitals, homes, schools, and transportation systems, they are vulnerable to both physical and digital threats. To mitigate these risks, backup systems and safety measures are essential.
Beyond the hardware aspects, the EU's AI Act requires high-risk AI systems to meet accessibility standards, while the EU Accessibility Act mandates that digital products and services, including AI-powered assistive devices, must be compatible with existing assistive technologies.
To facilitate the development and testing of these technologies, sandboxes and regulatory testbeds are needed for safe experimentation and formal compliance testing. This will help in ensuring the safety and efficacy of these systems before they are deployed.
The study group, organised by the Commission's DG CNECT, focuses on AI and robotics policy and the deployment of critical digital capacities. This group concerns data, AI value, and supply chains, including Eurostack, AI Continent Action Plan, AI Factories, EU InvestAI, and the EU Chips Act.
In addition to this, governments are funding projects for the development of modular robotic components, mobile robotics platforms, and open-source sensor systems. These initiatives aim to create a more integrated and accessible ecosystem for assistive technologies.
Moreover, the study group is also exploring the development of data spaces for accessibility-focused multimodal datasets, including low-resource languages, urban and indoor scenes, and training of specialised annotators with data accessibility expertise. This will help in creating more accurate and accessible AI systems.
Lastly, the search results do not provide information about governments that have optimized language processing learning models with compression techniques and specialized adaptations for low-energy, low-processing-capacity devices in hospitals, rural clinics, or home care settings. This is an area that requires more attention and investment.
In conclusion, the development and deployment of assistive AI technologies are a critical step towards making technology more accessible for everyone. With the right investments, policies, and infrastructure, we can make significant strides in this area, improving the lives of millions worldwide.
