Machine Learning & AI: A Platform for Growth

by Dimitrios Vlitas on Feb 22, 2019

Artificial Intelligence drives incredible amounts of economic value. Almost all of the economic benefit created by AI today is generated using the input-output data machine learning process (for more information read previous blog post). Standard examples of this are loan approvals and targeted online advertising.

In the first case, Machine Learning (ML) algorithms are fed with data describing client characteristics and outputs, informing of whether said clients are likely to be able to pay their debts or not, if accepted for a loan. With these data sets, machines learn from customers’ financial histories and become experts in evaluating new loan applications, thus significantly reducing risk for the bank.

Where online advertising is concerned, ML algorithms are provided with records of consumers’ habits, so they can easily identify customers with similar preferences and make appropriate suggestions for future purchases. A prime example would be the Amazon recommendation system. This allows companies to build more ”intimate” relations with their customers, benefiting both their bottom line and giving the customer exactly what they want.

Two success stories of employing ML within the banking industry are those of Man Group and Bank of America. Man Group Plc. in particular, built a system that evolved autonomously, finding money-making strategies where humans had failed to, and the results were startlingly good. By 2015, artificial intelligence was contributing roughly half the profits in one of Man’s biggest funds.

Consumers today are accustomed to the types of seamless mobile experiences provided by apps like Uber and Airbnb, and want banking experiences that match them in terms of ease, accessibility and speed. To this end, Bank of America introduced Erica, their new digital banking assistant. Erica uses artificial intelligence, predictive analytics and cognitive messaging to help customers make payments, check balances, save money and pay debt. As systems like Erica integrate deeply in our lives, we should expect the rise of fully-synchronised voice banking, payments and commerce.

iStock-682373850-colorer-low-1

 

The Rise of AI

There are three important factors that are responsible for the recent boost in all things AI. First, the availability of data. We are very good at producing and collecting data, generating enormous amounts of data on a daily basis, and as a result, we have had to develop ways to collect it more efficiently.

Secondly, our computational power has dramatically increased over the last decade. A graphic processing unit (GPU) together with a central processing unit (CPU) allow for parallel processing that greatly accelerates computing and its potential for further growth in the future.

The final factor, is the algorithmic innovation that has allowed us to process vast amounts of data with great scalability. These three factors allow us to leverage data for effective machine learning. Here lies one of the main causes responsible for the outstanding performance of ML algorithms. They can learn from massive amount of data, recognize patterns, combinations of patterns at a far greater speed than humans. This is the main reason that ML models can achieve results that surpass human level performance in many areas.

Besides the supervised methods of learning (input–output data), we have succeeded in transferring learning between domains. For instance, to go back to my original example of a  data set containing labelled images of either lemons or apples, we may have algorithms that have been trained there, to classify images of those two fruits.

We can then use this trained algorithm in areas such as medical diagnosis, in cases where we don’t have a lot of data. So we can use the learning acquired from the above data set of lemon and apple images, to distinguish adult x-ray images from those of children, for example.

Even though today we are experiencing an AI shared economy, with open resources and software available to everyone, it is the data that builds defensive business. Nowadays all the techniques necessary to train a ML model are widely accessible. We have enough access to computational power, so the only restrictive factor is data.

It is often the case that the existence of data drives the creation of products. As consumers use products, they create more data, that in turn is used to train machines yet further. In this manner, products can be continuously improved and the possibilities are endless. As a consequence, companies have access to enormous data sets, which becomes their greatest asset and the root of their growth.

It is commonly accepted that a company with a web page is not necessarily an internet company. Only when it posses the architectural organisational structure to leverage internet capabilities does it become an internet company. The same applies with AI. Just the use of ML algorithms does not guarantee a company’s best use of the new technology. The very first prerequisite is strategic data acquisition. Then, within and across the whole company, a unified data policy must be followed. As a result, every data science team and engineer could have access to data. After the collection of data, the next step is to build a centralised AI team, which communicates with all the other units in the company. Scientists from AI team then can address the individual needs of each unit.

AI is not a finished product. Is an autonomous self-learning process that has to be customised to fit the needs of the institution that makes use of it. It requires trained Data Scientists who can figure out the necessary input–output data, fine tune the ML algorithms and eventually produce well trained models.

It is necessary for a company to be alert and understand the abilities and limitations of AI. Then they should decide how AI could best meet its requirements. Throughout this process, certain products of pre-trained models can be helpful. The company must have a clear vision and the capacity to navigate its own way through the AI ecosystem. In this manner it remains ”lean” and not burdened by too many generic products provided by the preeminent vendor AI companies.

These companies have very skilled teams that may lead research, but in the end each institution has its unique characteristics and should adapt AI to them. Being ”lean” will enable these organisations to be more flexible and achieve exponential growth, much more quickly.

About
Dimitrios Vlitas
Dimitrios Vlitas

Head of Data Science at Strands, Dimitrios has more than 17 years' experience in the field of Mathematics and Computer Science and is Visiting Professor of Mathematics and Machine Learning at the University of Toronto.

Get the latest updates here

SUBSCRIBE HERE