The term ‘robot’ has come to be antonymic to ‘human’ and for obvious reasons. While in fiction they are frequently pitted against each other as opposing rivals, in truth artificial intelligence (AI) and human intelligence are actually more similar than they are different.
This is made clear by computing and AI expert Dr Ali Ahmed, Chief Data Scientist at NETSOL Technologies, who describes AI as “a field concerned with enabling machines to mimic human minds in learning and problem solving”. Dr Ahmed has a PhD in AI and signal processing and has studied imaging and computing for years at some of the top US institutions. Now back in Pakistan, he has established a new AI centre, is an advisor to the President on emerging technologies, as well as heading up NETSOL’s AI R&D division.
Dr Ahmed is actually not a fan of the term AI, which he describes as “a bit of a misnomer”. He admits that achieving human intelligence is a somewhat haughty goal that today’s technologies are far removed from. “This does not mean these technologies are currently useless,” he notes. “They are in fact tremendously useful in narrow contexts and have found many applications in almost every industry. Wherever there is data, you can hope to build models that act like humans.”
To do this, neural networks in deep machine learning imitate the brain’s ability to recognise data patterns. The real power of computers is to take the potential to spot patterns to a different level. “The language for us to communicate with computers is algorithms and statistical models. Once we tell a computer how to think like humans, it starts to look at data in a more intelligent fashion and discover patterns that are useful for performing a specific task,” explains Dr Ahmed.
AI in financial technology
AI is considered as an emerging technology for many industries. In healthcare, for example, it can analyse CT scans to predict cancers. But in some sectors, it has already been transformative. In the finance and leasing market, it has dramatically altered the competitive landscape, giving birth to financial technology (fintech) companies who, says Dr Ahmed, “use AI to gain competitive advantage over traditional banks.”
One of the main ways they achieve this is through enhanced customer service. “Today, most banks have a chatbot that is trained to answer clients’ questions. Customers might want to look at past transaction refunds, monitor recurring charges, see what their credit score is – they can ask all these questions to a chatbot and it will reply in a very intelligent manner within seconds.”
Another application is credit risk management, where banks can feed their vast amount of customer data into a machine learning (ML) algorithm, which then develops a deep understanding of customers’ financial profiles. “You can use this to assign credit scores and decide who to give a loan to. Credit scoring was previously performed manually. Because of this, there were many contradictions in decision-making; it was dependent on the agent to some extent. Using ML, banks are reducing non-performing loans up to 50% and boosting revenues significantly,” Dr Ahmed says.
He also expects targeted marketing campaigns to be developed using algorithms that predict customer churn. These will be highly customised; the goal is to offer the right product to convince an about-to-switch customer to stay.
Cybersecurity is another area benefiting from AI. “Cyberattacks are increasing,” says Dr Ahmed. “We have to stay one step ahead of them. AI looks at network traffic and is able to flag cyberattacks. It even identifies the attacks you haven’t explicitly programmed it to stop, by flagging events it hasn’t seen before.”
When you further consider fraud prevention and investment portfolio optimisation, the role of AI in financial institutions, and especially fintechs, is already clear. A recent Mordor Intelligence report valued AI in the fintech sector alone at $7.91bn last year, predicting it will reach $26.67bn by 2026. That’s a predicted CAGR of 23.17%.
One of the companies driving this boom is NETSOL Technologies, which provides software solutions to the global asset finance and leasing industry. The company has already integrated an AI-driven credit approval system into its Ascent platform, an adaptive platform designed to meet the specific needs of the global asset finance industry and has solutions for customised financial products and churn prediction. The company is also working on an upselling solution that uses AI to detect and inform customers when they can afford better products, using features such as expected income.
With increasing amounts of data being collected in a disciplined way, the road ahead leads to exciting places for AI. But there will be challenges too, primarily around the generalisability of AI models currently dependent on the hypotheses they are built for. Applying these same narrow models to new markets presents a challenge which Dr Ahmed describes as “being able to learn from little data – transferring previous knowledge and then requiring only a little bit of new data to start performing well.” Addressing this will be a true gamechanger for the future of AI.
