What are AI and ML Really? - And how can They Help Your Business?

Posted by Raiha Buchanan on 6/8/18 11:25 AM
Raiha Buchanan
Find me on:

Nowadays both Artificial Intelligence (AI) and Machine Learning (ML) are used interchangeably. Even though AI and ML are intrinsically related, to fully leverage the benefits for your business it is crucial to understand what they are. In this article we will give you the low-down of AI and ML:

What are AI and ML? What is the difference between AI and ML?

(Image credit: Lukas)

So, What is AI Then?

ML is one approach to achieve AI. AI is a machine´s capability to replicate behavioural activities that involve mental performance such as learning, reasoning & creating.  In other words, AI is a machine performing tasks that are characteristic of human intelligence (or another living being, e.g., an ant). This can include functions like planning, interpreting, problem-solving and understanding language. Currently, AI is being used to assess credit risk decisions, which were previously made by humans with paper applications; now this process is fully automated.

For example, in the financial industry, AI is being used to identify the most efficient way to get a customer to repay their loans. An AI system is given information about the reward, which in this case is a customer repaying, and a set of possible actions it can take for instance send an email. The system uses a trial and error approach to identify, which actions to take and with which individuals to achieve success.

Read here: Alternatives to Credit Scoring

Then, What about ML?

ML algorithms use programming and other computational methods to learn information directly from data. ML is a way to train an algorithm to identify and learn patterns in information. ML models are built without relying on a predetermined equation as a model, as more data points are provided the algorithms adapt to improve their performance. Natural patterns within data are identified to provide insight and enhance predictive capabilities. There are two main types of ML; supervised learning and unsupervised learning (see below).

Currently ML is used to predict the overall survival of patients with brain tumors, the likelihood of a product to be purchased based on its attributes, the expected clicks of advertisement, and to recommend running shoes.

What is the difference between AI and ML?

Typically within AI research communities and data science fields, in AI there is an agent (a computer) that interacts with an environment and learns from it, whereas in ML the agent is provided with data to train and learn from.

A  excellent example of this comes from Joacim Strand, a Data Scientist at Instantor. "Imagine teaching an agent to play Super Mario. In ML you would give the agent the data of people playing the game, and from this input, the agent can learn how to play the game and earn points. But in AI, the agent has to figure out how to play the game and earn points itself. In the AI scenario, the agent is given a set of possible actions it can take, e.g., jump, and a set of rewards, e.g., a point for every box it hits. The agent then tests these actions, in a trial and error fashion, it observes the outcomes of each action, and then uses this information to maximise rewards and to win the game."

The key difference between AI and ML is that AI has to learn information itself, it explores states, activities etc. itself.

What is supervised learning and unsupervised learning?

Supervised learning consists of leveraging previous knowledge to train a general model, which is used to make predictions about future events. For example, banks use the data they gather about the behaviour of their previous customers to make decisions about new customers. They have built general models that allow them to attempt to identify patterns in behaviour to determine ideal and non-ideal customers. Banks have learned through statistical analysis that information like salary, expenses, marital status and employment status of the borrower can help predict the creditworthiness of an individual.

At Instantor we have taken this one step further and have applied supervised learning techniques to an applicant's transaction history to make recommendations to financial institutions.

In contrast, unsupervised learning doesn't leverage existing data; instead, the algorithms need to identify patterns. Unsupervised learning consists of a statistical analysis of the underlying structure or distribution of data. Unlike supervised learning, in unsupervised learning data is not classified, there is no predetermined target for the model to learn. In unsupervised learning, the goal is to minimise a metric or objective. An application of unsupervised learning is grouping together similarly looking leaves to help identify plant species.

Unsupervised learning techniques are frequently used in companies to detect anomalies in data, for example, banks and lending companies use it to flag abnormal applications. To achieve this the model is given a series of credit applications; the model uses these credit applications to learn what average applications look like and what abnormal applications look like, and can thereby be used to detect fraud.

What are AI and ML? What is the difference?

Insight

Insight helps financial institutions to predict behaviours; to identify the likelihood of an individual repaying a loan and predict if an individual is likely to take on more credit soon.

Insight uses ML to analyse more than 90 predictive features and insightful patterns in historical banking behaviours, so we can help our customers to make tough calls easy,  and make more responsible risk decisions. Sharing deep market insights empowers our customers to reach out to their customer with the answers, they’re looking for. Instantor´s models are measured by Gini and the Area Under Curve (AUC), to ensure performance and predictive power.

The Economist’s Intelligence Unit (2018) identified the economic benefits of applied AI and ML across different sectors, including financial services. It points out an improved decision-making process, higher operating efficiency and better consumer experience at a lower risk.

Conclusion

Both AI and ML are already prevalent in the financial services sector. Both are already applied throughout the industry driving significant positive results. Instantor´s Insight is developed using predictive algorithms that empower decision-makers to make more accurate & profitable business decisions. If you would like to learn more about how our intelligent solutions can help your business, then book a session with us or try our free demo

Get a free demo today!

Topics: Product Updates, AI/ML