Narrow artificial intelligence: advantages, disadvantages, and the future of AI

Artificial intelligence (AI) is now embedded into our daily lives. Its applications have revolutionised whole industries and sectors, and it has similarly transformed the way people interact with technology, whether it’s using an email inbox at work, or choosing a new film to stream at home.

The technology for these applications is made possible through a type of AI called narrow artificial intelligence, also known as narrow AI or weak AI. 

What is narrow artificial intelligence?

Narrow artificial intelligence is what powers the AI applications most people are familiar with today, from streaming suggestions on Netflix to digital assistants such as Amazon’s Alexa and Apple’s Siri. It also includes technological developments such as facial recognition and driverless cars.

Narrow AI systems are designed and developed to perform particular tasks – ideally in much the same way a human being would perform the same task. They excel in specific areas, such as voice recognition, image classification, and data analytics, and are built using algorithms and machine learning techniques such as deep learning, neural networks, and validation.

The benefits of narrow AI

The rapid development and implementation of artificial intelligence during the 21st century has already demonstrated a number of advantages. For example, benefits of AI can include:

• Increased efficiency. Narrow AI has helped automate repetitive or mundane tasks. With these tasks now taken care of by computers, people have more free time for complex and creative activities. Automation, meanwhile, has increased productivity and reduced costs across several industries.
• Informed decision-making. AI algorithms can process vast amounts of data inputs, even big data, in short periods of time – sometimes almost instantly. For example, AI systems can quickly identify patterns or correlations in datasets that may have been missed during human analysis. 
• Helpful personal assistants. Virtual assistants such as Alexa, Siri, and Google Assistant use narrow AI technology to process commands. This means they can understand human language – and respond as needed. Through this technology, virtual assistants can perform tasks, answer questions, and even help organise people’s daily lives. 

Narrow AI has also made significant contributions to the healthcare sector globally. For example, AI-powered systems can aid in diagnosing diseases, analysing medical images, and personalising treatment plans – and by providing faster, more accurate diagnoses, this technology can help save lives.

The disadvantages of narrow AI

While narrow AI can be a powerful tool, it’s not without its drawbacks. These can include:

• Limited scope. Narrow AI systems are designed to excel in specific tasks, but lack the ability to adapt to new situations. They operate within predefined boundaries and may struggle when faced with tasks outside their expertise.
• Lack of human intelligence. Narrow AI lacks human-like intelligence, empathy, and common-sense reasoning competencies. This limitation restricts its ability to understand context, make nuanced decisions, or fully comprehend complex scenarios.
• Data dependence. Narrow AI systems rely heavily on high-quality and relevant datasets to learn and make accurate predictions. Biases and inaccuracies within datasets can result in flawed decision-making, or reinforce existing societal biases – leading to ethical concerns.
• Reduced human involvement. Over-reliance on narrow AI systems could potentially lead to reduced human intervention in decision-making processes. Most AI experts agree it’s important to find a balance between automation and human judgment to ensure the ethical and responsible use of AI technology.

Other types of AI

Artificial general intelligence (AGI)

Artificial general intelligence, or AGI, describes AI systems that possess human-level intelligence. These systems aim to exhibit the adaptability, learning capabilities, and reasoning abilities of a human being. AGI – sometimes called strong AI or general AI – however is still a hypothetical technology, and achieving AGI is an ongoing challenge in the field of AI research.

Superintelligence

Artificial superintelligence, or ASI, is another hypothetical area of artificial intelligence, one that aims for machine intelligence surpassing human intelligence, and possessing superior cognitive abilities. The concept of superintelligent AI remains largely speculative, but it does raise important questions about the issues associated with highly advanced AI systems.

Primarily these concerns relate to AI’s lack of human ethic and general incompatibility with our own morality and values and furthermore, the possibility of harm created by its capacity to act in ways that are uncontained or out of sync with our wants or needs.

Challenges, concerns, and risks around AI

New technologies and breakthroughs in artificial intelligence can be powerful tools, but as AI technology continues to evolve, so do the concerns around it. For example, there are:

• Ethical questions. AI raises ethical concerns around topics such as personal privacy, data security, biases, and considerations around things like black box AI. There are also more sinister applications of AI, such as deepfakes and the potential for autonomous weapon systems. 
• Job loss concerns. Automation driven by AI has the potential to disrupt various industries and result in job displacement. While it also creates new job opportunities, reskilling and upskilling programmes need to be implemented.
• Cybersecurity threats. Like all technology, AI systems can be vulnerable to cyber attacks and malicious use. And as AI technology becomes more sophisticated, so do the threats associated with it. 

One of the other frequently mentioned risks of AI in its current state is the huge amount of computing power required for AI apps and bots, which has significant knock-on environmental consequences:

“As big data, machine learning, and artificial intelligence continue to gain prominence in information technology, experts are raising concerns about the environmental costs of computation — primarily data and AI’s carbon footprint and greenhouse gas emissions,” Forbes reports, citing an MIT Technology Review that found the datasets and learning algorithms used to train AI models can emit more than 626,000 pounds of carbon dioxide equivalent.

Mitigating the risks posed by AI

Many of the challenges posed by AI – or challenges anticipated to emerge in the near future – are already being considered within the field. Some of the suggested solutions include:

• Creating AI ethics initiatives. Many businesses and organisations, such as IBM, and research institutes and governments are actively engaging in discussions around AI ethics and the real-world impact of AI. For example, initiatives such as the Institute of Electrical and Electronics Engineers (IEEE) Global Initiative on Ethics of Autonomous and Intelligent Systems aims to promote the responsible and ethical use and development of AI.
• Implementing regulatory frameworks. Many governments and policymakers are working on developing regulations, legislation, and other legal frameworks to ensure responsible AI development and deployment while also protecting user privacy and preventing bias. 
• Collaborating towards transparency. Encouraging and enabling collaboration between AI developers, providers, researchers, and other stakeholders can support transparency and accountability in AI systems. For example, sharing best practices, data, and methodologies can help address biases and promote responsible AI innovation.

Launch your career in the fast-growing field of AI

Join the AI revolution with the University of Wolverhampton’s 100%-online MSc Computer Science with Artificial Intelligence. This flexible Master’s degree has been developed for forward-thinking individuals who may not have a background in computer science.

You will develop a critical understanding of the processes, techniques, methodologies, and hands-on applications of artificial intelligence using contemporary software. You will also develop the communication, management, and organisational skills needed to lead data science projects, and to communicate findings within organisations, including to non-technical demographics. Topics of study include:

• data science
• machine learning
• intelligent agents
• data mining
• informatics