Three ways in which AI will change asset finance and asset-based lending.
Firstly, a definition of Artificial Intelligence ("AI") from the Oxford dictionary...
"The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. "
Could it be possible that tasks that currently require human intelligence within the asset finance space will be replaced by AI?
Autonomous Research, a global research firm (focused on financial services) predicted that over $1 trillion of today’s financial services costs could be replaced by AI.
I have picked three areas to analyse that have already started to shape the asset finance industry.
Before I begin, it is important to note that AI has many ethical and legal challenges, these challenges will continue to be the subject of much discussion. However, I have decided to only focus on the potential improvements with the knowledge that AI's application will be subject to law and ethical considerations.
What is the current state of play?
Recent advancements in computing power and the availability of cloud-based technologies has made the possible use of AI services for a few cents via an API.
Services such as Amazon SageMaker or Google Cloud AI are democratising AI by providing algorithmic tools that are now affordable to many asset finance businesses.
A few words of caution, are AI's current capabilities over-hyped? I believe the answer to that question is yes. However, it will change the way asset lending is transacted in the future.
Below are three use cases that will change the asset finance industry:
1. Fleet and Supplier Management
AI will allow greater scrutiny of asset reliability by developing a better understanding of asset breakdown and durability.
Using historic breakdown and accident data will enable fleet providers to predict the most likely breakdown rates. They will also be able to easily switch parameters to provide insight into how small changes to a customer’s leased fleet could yield better efficiency.
For transport assets this will give a statistical understanding of how an asset’s make and model play a part in the likelihood of breakdowns and faults.
Many firms collect breakdown histories of their leased assets, this can be used to dictate future procurement choices.
Fitting appropriate tracking devices that report back to the asset finance provider to ensure that their assets have not been taken abroad or stripped will enable technology to further protect their security. Tracking unusual movements of your asset could identify potential fraud.
Asset tracking will also move beyond simple monitoring of the movement of assets to intelligent decision making. In transport, monitoring driver behaviour, engine performance and fuel consumption will lead to a better understanding of how to protect assets from wear and tear. For example, this data can be used to incentivise a better driving style among lorry drivers. Not to mention the possibility of driver-less-vehicles making an impact on asset fleets.
Result: AI improves purchasing decisions and asset efficiency.
2. Automation and Data-Driven Decision Making
Decision making - AI will improve deal processing speed, give better predictions of yield and risk by using historic life cycle data from the loan book.
Beyond MS Excel models - Predictive modelling can be used to enhance decisions, however, there are three prerequisites for this to work. First, you will need the availability of data. To this end, many asset finance firms are sitting on a gold mine of data from origination to end-of-life that they never intend to use in this way.
Second, you need to develop the ability to train and tune machine learning algorithms at scale. This requires an investment in data science expertise.
Finally, you need an understanding of how to turn insight into action by:
- predicting portfolio outcomes,
- improving scorecards,
- improving underwriting decisions,
- giving a better understanding of risk & reward and
- quicker AML and KYC processing
For those wanting even more advanced techniques, neural networks could be used to improve results even further.
Fraud detection - According to The Finance & Leasing Association (FLA); counterfeiting, asset stripping, carousel fraud, cybersecurity, identity theft and client insolvency are the most common types of asset fraud.
Machine learning could be used to detect unusual behaviour and patterns in data, whether that data is about assets or customers. For example, what are the common data patterns that can be detected in the fraudulent acquisition of assets?
The most interesting AI development is the combination of computer science and behavioural science. This combines our understanding of human behaviour with historical datasets and computing power.
AI will use this combination to enhance fraud detection models providing better early warning systems. As an example, Paypal’s Hui Wang, Director of Risk Sciences commented: “We take trust very seriously. It’s our brand. We have to decide in a couple of hundred milliseconds whether this is a good person, [in which case] we will give him or her the best and the fastest and the most convenient experience. Or is it a potentially bad guy and we have to insert some friction?”
Regulatory compliance - Regulated deals could be subjected to automated regulatory compliance checking. Internal business rules regarding sector/client/risk exposures could move from monthly/quarterly checks to near real-time.
Result: AI becomes an essential tool for lowering costs, measuring risk and creating opportunity.
3. Document processing
Most asset lenders process a large amount of documentation per deal. The vast majority of these documents are typically pdfs or images.
It is true that pdf document readers have been in place for a while now. What has changed is the increasing AI based computing resources.
Natural Language Processing (“NLP” another AI technique) is enhancing our ability to classify and designate documents by providing meaning that was only possible through human analysis.
Tools such as Google Vision API will continue to make inroads into the automated processing of documents and images. The Google Vision API and its competitors will allow the mass categorisation and processing of thousands of invoices, images, ownership documentation and KYC documentation at scale. This could include directly interface with accounting systems for automatic filing of purchased assets.
Fraudulent documents could include false invoices, receipts and bank statements used to purchase an asset. Asset re-finance and sale & lease-back deals are particularly vulnerable. AI could use simple rules to detect anomalies in asset deal documentation. For example, checking if an asset supplier exists and if a document conforms (formatting) to previous documents from that supplier.
This change could also be extended to improve the user experience. For example, NLP could be used to sort and filter customer emails that are showing signs of distress (through sentiment analysis) providing more effective triage to the right internal customer service team member.
Result: AI helps to process documents and detect fraud.
To conclude…
With all this hype – it's easy to say that AI will replace human interaction. This will not be the case, as I see AI as an additional tool. If man and machine could collaborate effectively this will drive up productivity and efficiency of the asset finance sector.
Helping Startups with Business, Data, App, & Tech.
1 年Carl, thanks for sharing!
Application Security Software | Synopsys
1 年Carl, keep it up!
Talks About - Business Transformation, Software Solutions, Organisational Change, Business Efficiency, Sales, Scalability & Growth
2 年Great post?Carl, thanks for sharing!