Recently, Artificial Intelligence (AI) under Fourth Industrial Revolution is driving major paradigm shifts across all sectors and countries. This technology has seeped into our everyday lives, transforming how we live, work and entertain ourselves. This has led to a focus on the US and China, being key players, but not the only ones, in the AI race. This is because of the first comer advantage of AI technology. The premier for this technology will guide the rest of the narrative. There is no doubt on the power and generational impact of AI that will affect not only national economies but society on the whole. A report by PWC estimates a global increase of $16 trillion in GDP by 2030 in lieu of AI technology.
In addition to the financial and societal benefits, AI raises significant questions for society, economy and the government setups. Is there an opportunity for international collaboration in fast developing technologies like AI to achieve global benefits? Hence, a dire need for formal national AI frameworks and international standards for controversial aspects of AI such as data and algorithmic bias, transparency and ethical use for decision-making to steer this shift where it benefits the most.
AI has the potential to improve lives of millions given the use is inclusive and ethical. However, this requires proper planning and oversight by policy makers. It can either promise to solve or worsen complex and pressing problems like inequality and gender bias. The AI strategy needs to focus on both the private sector led economic growth and the added benefit of social inclusion. This is encouraging policy makers and private sector to take crucial steps to win the digital race. The policy should attempt to minimize risks and maximise benefits for the whole society. Therefore, to address different needs and opportunities, we need to understand what is AI and its types.
AI is broadly used as a blanket term. A widely accepted definition by Russell and Peter, at its root define AI as ‘the ability to replicate or imitate human thought’ and more recently as ‘using human reasoning as a model to solve problems.’ However, previously terms like manufacturing were used for the production of everything from cars to clothes to processed foods. Thus, it is more appropriate and useful to define AI through its intended purpose.
At present, there are at least 8 general types of intent AI tech is used for:
|1||Transactional||To sell||Online sales and customer service.|
|2||Communicative||To inform||Maps, restaurants, suggestions, cultural and social interactions|
|3||Educative||To learn||Teaching assistants, online courses and special education and leaning|
|4||Social||To connect||Social network, identity authorization to personalize recommendations|
|5||Play||To entertain||Online gaming and smart toys|
|6||Administrative||To manage||Finances and schedules through digital integration into our daily lives|
|7||Diagnostic||To identify the abnormalities||Used in health industry for medical symptoms and used in security surveillance security|
|8||Behavioural||To change and modify the human behaviour||Examples include exercise, changing mental health and eating habits to improve overall well-being|
For a dynamic AI system, it needs to be combined with data related to user’s underlying condition and preference. These usage patterns can drive algorithms for user friendly and user-centric AI products and services. This is a chicken and egg situation as more data allows the algorithm to improve and allow for personalized experience that has better outcomes for sales and customer retention.
On the other hand, better smart products and services are used widely, thus, generating enormous amounts of data to be used by the self-learning algorithms. This is one of the main reasons’ most of the developers build, deploy and then tweak as deemed relevant after looking at the response. Most of the AI systems are a combination of being trained and self-learning. They initiate the process using the given data and then from there onwards use the data that is constantly generated with usage to adapt to user preferences.
Thus, data is the driving force behind AI. AI advances are more likely in countries with open data sources and that promote data sharing. The AI strategy needs the public and private data to be centrally collected and made publicly available. Though, a huge task with open data sources is to balance the need for data privacy. There is clear evidence on the dangers of releasing a self-learning chatbot in the public arena without any human input. A famous example is of the Microsoft’s chatbot, Tay as it learned hate speech within hours of its launch. This is a reminder that AI is based on statistics and not the belief system. Thus, hybrid systems are more beneficial to become a norm in behavioural and diagnostic AI to minimize risk.
Moreover, continuous research and innovation is essential to growing customer confidence and the integration of these systems in our daily lives. The clarity of purpose is reflected how engagement and outcomes are measured and to ensure a framework for ethical integrity. Currently, the most common purpose of AI is transactional. The winners for overall conversational AI will be those who show meaningful use and beneficial outcomes that serve the purpose of their systems.
Conclusively, AI has exciting and promising potential to augment the everyday life. Though, the greatest popularity will come from transparency in practice, measuring meaningful outcome and fair monetization. Moreover, AI faces strong criticism based on the ethical aspects, thus, a lot of questions surrounding how policymakers address issues related to privacy and moral use of data. Policy makers need a legal framework that provides concise and comprehensive guidelines for all stakeholders to enable broad yet safe use of AI systems. The policy needs to act as a roadmap to tackle issues as they arise and promote using AI for broader socio-economic benefit.