Our Work on Building safeAI Systems for Businesses and Society
Artificial Intelligence (AI) systems are being used by many industries and also inour lives.
AI systems are influential across the industries in intelligently automatingmany processes and creating savings and efficiencies. Many AI products, such ashome robots or intelligent software on our mobile devices are increasinglybecoming part of our daily activities. While AI brings such immense help, the risksassociated with this complicated technology is also well recognised.
There are numerous attempts for governance of AI to reduce their risks inbusiness or in the society. However, these are all local, limited by human efforts(and errors) and dispersed.The proposed project, safeAI, aims to contribute to minimising risks of AI bypromoting the ways in which safe and reliable AI can be developed mymanufacturers of AI products and also safe AI approaches can be adopted by theusers of AI systems.
SafeAI does this through an intelligent and automated systemto provide customised information based services to clients according to theirneeds.In this was, it revolutionises the man made, limited and local versions of AI risksregulations, guidelines and practices for producing safe and reliable AI products orservices.
AI risks in the industry that need to be mitigated:
- Regulatory risks: AI regulations are being developed manually (subject tohuman error) and slowly, and they are insufficient to keep up with fast changesin AI industry.
- Data risks: Lack of right amount and representative data to work on createsevere biases in AI systems — underserved communities around the world arebeing discriminated.
- Software risks: There is a lack of commonly agreed or accumulated globalknowledge on AI software development processes –they are black boxes.
- Consequential risks: Potential harm as and after being in use — a self-driving carrecently killed an elderly lady, face recognition systems have surveillance andprivacy risks.
- Unknown risks: as AI industry, with IoT and Cloud, evolves fast and becomesmore complex, so does the AI risks.
Covid-19 & AI: Fighting Pandemics
It is so obvious that using technology such as AI, can help with epidemics in similar circular phases as disaster management – mitigation or prevention, preparing to fight, responding or fighting the epidemic, and finally recovering from the impact.