We all know about machine learning when it comes to Japanese droids or Rhoomba intelligent vacuum cleaners, but how is machine learning being used in finance and fintech? As you will discover, the use of machine learning is both prolific and amazing. We will soon look back and wonder how we lived without machine learning.
#1 Fraud protection and fraud prevention
“Machine learning will automate jobs that most people thought could only be done by people.” – Dave Waters
The brilliant way that machine learning has been implemented to help protect against fraud is amazing when you consider the sheer weight of staff/human time required to do the same job. Machine learning can pick up on several factors all at once and put them together to figure out which threat is where, and then guard against that threat with something as simple as an access denial or the removal of vital information.
Malware is more advanced
This is especially important these days when malware doesn’t simply load onto computers but is instead split up and cast around the Internet, where victims pick up pieces like they were part of the malware puzzle. New threats are identified by the thousand every month, and hacking attempts in the US alone are blocked at a rate of millions per day. Machine learning is so sophisticated that it can block requests before they even reach US networks.
Governments use machine learning
“Machine Learning: A computer is able to learn from experience without being specifically programmed.”- SupplyChainToday.com
The US government receives at least one million cracking or hacking attempts per day from China alone, and they would be successful if it were not for machine learning perpetually countering their attempts and learning from it. Successful hacking and cracking reached its peak in 2011, and has slowly gone down each year thanks to advanced machine learning (the 2014 Fappening doesn’t count as it was not a case of hacking, it was phishing).
#2 Automated financing
Algorithmic trading is already quite common, albeit open to manipulation as we have seen with newer cryptocurrency markets. Automated financing is also seeing a surge within finance and fintech circles. What is happening is impressive to say the least.
The problem is human error
We have all seen the great things peer-to-peer lenders are doing without the help of banks, but they are all powered by humans. Each element, even basic online applications, are administered, managed and overseen by humans. This leaves the system open to attack, open to human error, and makes the system very slow.
Some companies are using machine learning
There are companies like Leads Market that have integrated automated financing to such a level that the machine learning systems can anticipate a lender and a publisher’s needs with the submission of just a small amount of information. Everything from the quality control process to precision lead generation is streamlined thanks to innovative machine learning. This, along with continuous follow ups with clients, is how the company is able to operate so efficiently without having to rely on having a massive staff team.
#3 Banking security
Oddly enough, it is easier to research details into how the US and UK governments use machine learning to guard against threats than it is to find a bank that will spill the beans. Though we all expect that they use machine learning for their security, there is also a notable increase in the amount of machine learning involved with the money transfer network. Banks are exploiting big data that they are pulling from their use of money-transfer networks and are using machine learning to identify possible fraud threats.
How banks detect fraud
It is not just about which country the money is going to, or which company is having fraud reports issued about it. They are even able to use demographics, though it is not clear how since they will not release information about the personal data they are using.
Wire transfers are changing
Credit cards have been using machine learning to monitor fraud possibilities for years, but banks are focusing more on wire transfer, which are typical avenues for fraud because they send money quickly and are very difficult to reverse. The next time you send a wire transfer and suddenly the transfer is frozen for two working days, you may have machine learning to blame.
#4 Automated marketing
We all know how machine learning has affected automated marketing when it comes to email marketing and chatbot marketing. They have made it so people can think they are having a conversation with a real person. Reality is that they are really talking to a very well-programmed machine. Yet, it is only recently that machine learning has entered the world of free website promotional methods such as use in social media, on forums, blog comments and so forth.
According to Larry Page –
“Artificial intelligence would be the ultimate version of Google. The ultimate search engine that would understand everything on the web. It would understand exactly what you wanted, and it would give you the right thing. We’re nowhere near doing that now. However, we can get incrementally closer to that, and that is basically what we work on.”
More than just a spam bot
Wait a minute, you cry. Spambots have existed for years. Automated social media is almost a bedrock of many online marker’s free promotional methods. Although this is true, automated systems (like Buffer, for example) still require epic amounts of user input. Simply because social media, forums, Internet servers, government groups, and spam catching software have become very good at spotting auto-generated content.
Countering safeguards set up by social media
High-end machine learning means a program can identify the safeguards that social media providers set out. And counter them with the publication of thought-out material that is unique to each profile, or group, or even to each comment section. This level of machine learning is not cheap. But it can essentially automate the promotion of a website on places like social media. While avoiding having social media accounts banned or marked as spammers.
#5 Insurance claims and fraud
Put in simple terms, insurance companies are entering the details of a claim into the computer. The computer then runs a comparison against all other claims and their outcomes. It then gives a reading on what the claim outcome is likely to be. This isn’t very impressive at first. However, an unexpected benefit is that it is able to identify factors that indicate that fraud may be occurring.
Probability becomes a warning system
Obvious factors are taken into account by the software. Including how long the claimant has had an account with the company. And if the claimant has a lot of debt and the details of the incident itself. It even takes smaller things into account like the account holder’s age, gender, policy changes and so forth. By taking this information into account, it can compare said factors quickly. It will then generate a percentage warning that suggests the claim may be part of some type of fraud.
Final Thoughts: know the limitations of machine learning
Many foolish people claim that machine learning is the key to predicting the future with absolute certainty. They also believe that bitcoin will rise, and they can learn how to manipulate the stock market. It is just not possible. The philosopher Nassim Nicholas Taleb wrote the book “Black Swan” in which not only did he confirm the chaos theory, but he proved by their very nature that any prediction is wrong because the past cannot be used to predict the future.
Nassim goes on to show how people use the information to predict the future. He also explains how their confirmation biases make them think they are correct. Even if machine learning could calculate every single variable, it wouldn’t matter. Simply because of what Nassim proved and because of the Chaos theory. Plus, the more you rely on such methods, the harder you fall when they fail.
In short, machine learning has a lot of uses. We’re going to see its adoption across far more sectors than we imagine. However, it has limitations as well. This means some of the most exciting machine learning prospects (those involved in prediction) are ironically the most useless.