What is AI and how is it being used?

Artificial intelligence (AI) is a broad term that has been around for decades with the concept dating back as far as the Dartmouth Conferences of 1956. How we interpret AI has changed over the years and has become the all-encompassing term for machine/algorithmic capabilities that imitate human behaviour. It imitates how the human brain works and interprets data to make decisions.

Speech recognition technology powered by AI now, in certain cases, has a similar ability and efficiency to the human brain. When it comes to image recognition, AI is now outstripping humans and recognising patterns that we ourselves cannot perceive. Our human brains can only keep track of so much information at a time, but AI does not have that constraint. However, AI is nowhere near being able to overtake the capabilities of the human brain. AI in its present form is focused on specific tasks rather than the generic machine that is the human brain.

Narrow or Weak AI

Narrow, or weak, AI is the area most successfully achieved so far. It is good at performing a single task. Simple statements can be combined into rules that mimic human abilities. Narrow AI has been successful in customer acquisition, profiling, and data mining.

Machine Learning

Machine learning AI can modify itself when exposed to more data without human intervention. It "learns" by attempting to optimise algorithms to achieve a predefined objective, minimising errors and maximising success. Early examples of this included machines learning to play chess.

Deep Learning

Deep learning is a subset of machine learning which takes learning to a much higher level with higher accuracy and much more hardware and/or training time. Deep learning has a higher number of layers of artificial neural networks and requires substantial computing power that has only become feasible in industry this last decade. Deep learning performs well on machine perception tasks involving unstructured data such as images or text.

Artificial General Intelligence or Strong AI

General AI is otherwise known as human-level or strong AI. It is a machine that can perform any intellectual task that a human can. It can reason, plan, solve problems, think abstractly, understand complex ideas, and learn quickly from experience. Whilst some people believe this type of technology is right around the corner, the reality is that most experts believe we are decades away from achieving this level of AI.

Narrow AI has been applied to almost all spheres of life and all business sectors. However, some sectors have really embraced AI and have seen its use transforming their markets. Some examples include:


The manufacturing sector has been using sensors for decades, but now the data that is being collected can be processed by AI.

AI can make predictions out of big data to predict machine downtime in a way that the human brain cannot. The human brain is unable to process the huge quantity of data quickly enough, nor is it able to capture micro signals that an IoT connected AI system can. AI is also starting to be deployed to predict and avoid disruption to supply chains, analysing geopolitical and weather data, as well as helping design and monitor maintenance plans to make them more efficient.

In addition, we are starting to see equipment-as-a-service business models emerge from Original Equipment Manufacturers (OEMs). AI coupled with insurance enables OEMs to get more comfortable with the new risks that this new business model can bring.


Automated vehicles are perhaps the most prominent example of how AI is being used. In 2005 we had the view that only humans could analyse images and react at high enough speed to be able to operate a vehicle safely. However, now we have very real examples of AI algorithms being used to solve this very real everyday problem. By 2035 there are expected to be over 21m self-driving cars on the world's roads.

At the moment, there are still no autonomous vehicles that can legally operate without a human driver as there must be a human operator to react to any safety warnings. However, the market is making fast progress towards full automation.

In the larger mobility space, we see a lot of potential for the use of AI. Besides the typical scenarios around route planning and navigation, we are starting to see efforts in the domain of "Mobility as a Service". AI can play a useful role in simplifying and managing the day to day commute of an individual through predictive routing, shared service allocation etc. Indeed, we are seeing start-ups emerge in this area which have carved out a niche for themselves.

Financial services

AI is playing an increasingly large role in decision making in the financial services space. Algorithmic credit scores are increasingly important when it comes to judging an individual's creditworthiness.

In the insurance sector, AI algorithms are also being deployed at large scale to enable faster claims response rates as well as detecting fraud. Algorithmic trading is also carried out on Wall Street, with bots carrying out stock buys and sells. However, financial services regulations are restricting how much the technology can be used until more is known.


One of the biggest benefits from AI has been the ability of the tech to leapfrog infrastructural and skillset gaps in the healthcare industry. AI is being used to offer affordable healthcare in remote areas by playing a triage role, considerably accelerating the growth of the healthcare industry into new markets.

In private healthcare markets it is also being used to determine whether or not requests to see a specialist or health insurance claims are approved. Computer-aided diagnostics can also help physicians interpret x-ray, MRI and ultrasound data more rapidly.

Silicon Valley entrepreneur and investor Vinod Khosla believes that up to 99% of medical needs can be solved through AI, big data and improved medical software and diagnostics. One example of AI in action is this space is the partnership between Google's DeepMind and Moorfields Eye Hospital to develop AI applications for healthcare, including analysing eye scans for early signs leading to blindness.


​​Tyranny of Data

Join our next webinar "Tyranny of Data: A Technologist and Cyber Risk Engineer debate the dangers of AI" on 10 December

​What are the emerging risks of AI?

​What can risk managers do to try and mitigate some of the risks posed by AI?