Connect with us

Technology

The pros, cons, and best practices when it comes to using generative AI

Colin Redbond, SVP Product Management, SS&C Blue Prism

We’ve all heard about growing interest in generative AI, especially following the release of ChatGPT. The artificial intelligence (AI) large language model, and ones like it, have the potential to revolutionize business operations and accelerate digital transformation journeys. Those that get ahead of this trend are set to gain a significant competitive advantage, with generative AI showing no signs of disappearing anytime soon.

While historically, AI projects have been long, expensive, and complex, generative AI has the potential to reduce time to value for digital transformation initiatives and make advanced technologies more accessible to a greater cross section of people thanks to its ease of use and learning capabilities.

How generative AI supports digital transformation

Digital transformation initiatives went into hyperdrive following the pandemic, but most organizations have yet to maximize business outcomes with their current automation plans. Intelligent automation combines technologies, like generative AI, robotic process automation (RPA), business process management, etc., to reengineer processes and drive business outcomes.

So, what’s the value of generative AI? Its range of capabilities and accessibility is unprecedented, marking an exciting time for the automation space and any sector standing to benefit from advancements in natural language processing, including healthcare, finance, and customer service. However, generative AI is still limited, in a sense, to its own domain knowledge. 

When you combine its unique capabilities with the power of intelligent automation, the impacts for digitalization are extraordinary. Generative AI can be used to automate tasks that were previously only possible for humans to perform, such as generating new marketing copy, designing new product prototypes, or creating personalized content for each customer. It can suggest automations and enable a greater cross section of workers to initiate the development of automations thanks to its ease of use. Automations can then be designed within designated governance parameters and best practices.

By automating these tasks, employees can reduce their workload, supporting work-life balance, while also increasing their efficiency, reducing company-wide costs, and improving the accuracy and quality of their output. Employees work with generative AI to deliver superior results.

When it comes to creative work, humans add color and empathy, which technology can only try to mimic. Generative AI gives them a starting point, helps with idea generation, etc. Human workers provide their uniquely human abilities to read between the lines and their emotional empathy for which AI is not substitute.

For many people, when they think “generative AI,” they think about written content or even AI art, but the use cases for generative AI relate to the day-to-day operations of most office workers.

For example, automated emails can exhibit a greater degree of personalization and improve resolution times. For more complex or high-level emails, generative AI can be used to draft an adequate, personalized email, with all needed information, and a human can then review and tweak if needed.

A similar process can unfold with contact center processes, generative AI bots can progress customer communications significantly before needing to loop in a human employee – if they need to be looped in at all for simple enquiries. This ensures human employees’ time is used effectively and as many customers as possible are being serviced, especially with generative AI bots being able to work around the clock. Error handling is improved with error messages providing context that enables immediate resolution.

Intelligent document processing (IDP) solutions, which use a combination of optical character recognition and AI to extract information that is locked away in documents, are enhanced by generative AI capabilities. This is especially important for financial and healthcare services. Generative AI’s understanding and learning features better equip it to contend with unstructured data, an area that has been a weak point for IDP solutions, which have been confined to structured and semi-structured documents at best.

Generative AI can also help improve the overall performance of intelligent automation systems by allowing them to adapt and learn over time by analyzing the results of previous tasks and using that data to generate new content or output.

The need for governance and risk management to unlock the potential of AI

However, businesses need to make certain considerations before they explore adding generative AI to their toolkit to accelerate digital transformation, since its outputs can have a significant impact on a company’s reputation, revenue, and legal liabilities. A clearly defined corporate governance risk management strategy and set of operating principles around this need to be developed. Done right, generative AI can support an automation strategy that is even more innovative, cost-effective, and productive than anything we have seen before.

Reasons why governance and risk management considerations are important when using generative AI:

  • Help ensure AI-generated content does not violate intellectual property, privacy, or other laws
  • Make sure use of generative AI aligns with your organization’s ethical principles
  • Maintain your organization’s quality standards and confirm outputs are consistent with expectations
  • Ensure the right information is used for the right purposes to protect sensitive information and privacy

How to develop a governance and risk management strategy

A clearly defined strategy and concomitant operating principles maximize the benefits of generative AI while mitigating any outfall. Developing a strategy involves several key steps:

  1. Define the scope: This includes the types of content you will be generating, the data you will be using, and the intended use cases for the content. This helps with identifying the specific risks and governance requirements that apply to your initiatives.
  2. Identify risks: These may include legal risks such as infringing on intellectual property, ethical risks such as bias in generated content, and security risks such as the potential for data breaches. You may need to engage with legal and compliance experts to identify all potential risks.
  3. Establish governance requirements: Based on the risks you’ve identified, establish governance requirements that will mitigate those risks. These may include policies and procedures for data handling, content review, and compliance with regulations.
  4. Develop a risk management plan: Outline how your organization will mitigate and manage risks. This may include risk assessments, monitoring, and regular reviews of governance practices, as well as processes for identifying and addressing any issues that arise.
  5. Train employees: It’s important to train employees on governance and risk management practices. This may include training on data handling, content review, and compliance with regulations. Make sure all employees who will be working with generative AI understand the risks and their responsibilities for mitigating those risks.
  6. Monitor and review: Monitor and review your governance and risk management practices on an ongoing basis. This will help you identify any gaps or issues that need to be addressed and ensure that your practices remain effective over time.

Like all advanced technologies, generative AI’s impact is positive – so long as you take the steps necessary to ensure you’re using it the right way. There’s no turning back the train, generative AI is here to stay – full steam ahead. The best approach is going to be to embrace it with care and work with providers when it comes to decision making around implementation. The possibilities behind generative AI are exciting – so let’s work to get it right and make it a force for good.

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Business

Empowering banks to protect consumers: The impact of the APP Fraud mandate

Source: Finance Derivative

Thara Brooks, Market Specialist, Fraud, Financial Crime & Compliance at FIS

On the 7th October last year, the APP (Authorised Push Payment) fraud reimbursement mandate came into effect in the UK. The mandate aims to protect consumers, but it has already come under immense scrutiny, receiving both support and criticism from all market sectors. But what does it mean for banks and their customers?

Fraud has become a growing concern for the UK banking system and its consumers. According to the ICAEW, the total value of UK fraud stood at £2.3bn in 2023, a 104% increase since 2022, with estimates that the evolution of AI will lead to even bigger challenges. As the IMF points out, greater digitalisation brings greater vulnerabilities, at a time when half of UK consumers are already “obsessed” with checking their banking apps and balances.

These concerns have contributed to the implementation of the PSR’s (Payment Systems Regulator) APP fraud mandate, which was implemented to reimburse the victims of APP fraud. APP fraud occurs when somebody is tricked into authorising a payment from their own bank account. Unlike more traditional fraud, such as payments made from a stolen bank card, APP fraud previously fell outside the scope of conventional fraud protection, as the transaction is technically “authorised” by the victim.

The £85,000 Debate: A controversial adjustment

The regulatory framework for the APP fraud mandate was initially introduced in May 2022. The maximum level of mandatory reimbursement was originally set at £415,000 per claim. The PSR significantly reduced the maximum reimbursement value to £85,000 when the mandate came into effect, however, causing widespread controversy.

According to the PSR, the updated cap will see over 99% of claims (by volume) being covered, with an October review highlighting just 18 instances of people being scammed for more than £415,000, and 411 instances of more than £85,000, from a total of over 250,000 cases throughout 2023. “Almost all high value scams are made up of multiple smaller transactions,” the PSR explains, “reducing the effectiveness of transaction limits as a tool to manage exposure.”

The reduced cap makes a big difference on multiple levels. For financial institutions and payment service providers (PSPs), the lower limit means they’re less exposed to high-value claims. The reduced exposure to unlimited high-value claims has the potential to lower compliance and operational costs, while the £85,000 cap aligns with the Financial Services Compensation Scheme (FSCS) threshold, creating broader consistency across financial redress schemes.

There are naturally downsides to the lower limit, with critics highlighting significant financial shortfalls for victims of high-value fraud. The lower cap may reduce public confidence in the financial system’s ability to protect against fraud, particularly for those handling large sums of money, while small businesses, many of which often deal with large transaction amounts, may find the cap insufficient to cover losses.

The impact on PSPs and their customers

With PSPs responsible for APP fraud reimbursement, institutions need to take the next step when it comes to fraud detection and prevention to minimise exposure to claims within the £85,000 cap. Customers of all types are likely to benefit from more robust security as a result.

The Financial Conduct Authority’s (FCA’s) recommendations include strengthening controls during onboarding, improving transaction monitoring to detect suspicious activity, and optimising reporting mechanisms to enable swift action. Such controls are largely in line with the PSR’s own recommendations, with the institution setting out a number of steps in its final policy statement in December 2023 to mitigate APP scam risks.

These include setting appropriate transaction limits, improving ‘know your customer’ controls, strengthening transaction-monitoring systems and stopping or freezing payments that PSPs consider to be suspicious for further investigation.

All these measures will invariably improve consumer experience, increasing customers’ confidence to transact online safely, as well as giving them peace of mind with quicker reimbursement in case things go awry.

Going beyond the APP fraud mandate

If the PSR’s mandate can steer financial institutions towards implementing more robust security practices, it can only be a good thing. It’s not the only tool that’s shaping the financial security landscape, however.

In October 2024, the UK government introduced new legislation granting banks enhanced powers to combat fraud. An optional £100 excess on fraud claims has been introduced to encourage customer caution and combat moral hazards, while the Treasury has strengthened prevention measures by handing out new powers to high street banks to delay and investigate payments suspected of being fraudulent by 3 days. The extended processing time for suspicious payments may lead to delays in legitimate transactions, making transparent communication and robust safeguards essential to maintain consumer trust.

Further collaborative efforts, such as Meta’s partnership with UK banks through the Fraud Intelligence Reciprocal Exchange (FIRE) program, can also aid the fight against fraud. Thanks to direct intelligence sharing between financial institutions and the world’s biggest social media platform, FIRE enhances the detection and removal of fraudulent accounts across platforms such as Facebook and Instagram, not only disrupting scam operations, but also fostering a safer digital environment for users. The early stages of the pilot have led to action against thousands of scammer-operated accounts, with approximately 20,000 accounts removed based on shared data.

Additionally, education and awareness are crucial measures to protect consumers against APP fraud. Several high street banks have upgraded their banking channels to share timely content about the signs of potential scams, with increased public awareness helping consumers identify and avoid fraudulent schemes.

Improvements in policing strategies are also significantly contributing to the mitigation of APP fraud. Specialized fraud units within police forces have enhanced the precision and efficiency of investigations. The City of London Police and the National Fraud Intelligence Bureau are upgrading the technology for Action Fraud, providing victims with a more accessible and customer-friendly service. Collaborative efforts among police, banks, and telecommunications firms, exemplified by the work of the Dedicated Card and Payment Crime Unit (DCPCU), have enabled the swift exchange of information, facilitating the prompt apprehension of scammers.

How AI is expected to change the landscape

The coming months will be critical in assessing these changes, as institutions, businesses and the UK government work together to shape security against fraud in the ever-changing world of finance.

While fraud is a terrifyingly big business, it’s only likely to increase with the evolution of AI, making it even more critical that such changes are effective. According to PwC, “There is a real risk that hard-fought improvements in fraud defences could be undone if the right measures are not put in place to defend against fraud in an AI-enabled world.”

Chatbots can be used as part of phishing scams, for example, and AI systems can already read text and reproduce sampled voices, making it possible to send messages from “relatives” whose voices have been spoofed in a similar manner to deepfakes.

Along with other innovations, tools and collaborations, however, the APP fraud mandate, UK legislation and FIRE can all contribute towards redressing such technological advances. Together, this can give financial institutions a much-needed boost in the fight against fraud, providing a more secure future for customers.

Continue Reading

Business

AI and Data Interoperability are Crucial for Success in the Financial Industry

Source: Finance Derivative

Written by Yohan Lobo, Senior Industry Solutions Manager, Financial Services at M-Files.

Businesses within the financial services sector are among the  industries leading the way in delivering AI initiatives to enhance services and improve decision-making however, rich data and strong infrastructure is the essential foundation for successful implementation.

Still plagued by inefficient manual processes and lack sufficient data resources, only 31% of organisations are on track with AI integration. Models that operate with AI are only as good as the data we feed into them so firms need an optimised system that can handle the high volumes of client and business data.

Financial institutions should address these gaps by investing in a robust data infrastructure that connects these siloed sources, creating a firm foundation on which they can build new AI initiatives.

The Pitfalls of Unorganised Data

Financial information can often be scattered across various locations in a range of formats, such as market insight presentations analysis, underwriting documents, or client emails. Without a predefined format, this disconnected data makes it challenging for AI systems to interpret information effectively and delivers inaccurate analytics that could take the business in the wrong direction.

Many institutions need help organising documents generated across disconnected systems and stored in duplicate data stacks which may produce conflicting versions of the truth. In a sector where client relationships are built on trust, responding to these data issues using obsolete tools, disrupted workflows, and any misstep in data consistency could lead to reputational damage, financial loss, or regulatory fines.

Organizing financial data like transaction records, customer data, and financial reports under centralised and labelled repositories, will make data collection and analysis for projects more accessible.

With these data management tools, firms can automate the process of organizing unstructured data that is easy to find, store and use. This can liberate their teams from the drudgery of manual processes while eliminating the potential for human error, resulting in richer data sources that is ready to fuel AI powered productivity.

Demystifying AI into a Workforce Ally

Workforce preparation and readiness may be an underestimated aspect of AI business readiness employees might be sceptical of AI accuracy and capability based on anecdotal stories following the failed usage of this new technology.

In 2023, a US attorney found himself embroiled in an AI disaster after using an AI chatbot to research precedents in a lawsuit against Colombian airline Avianca. In this case false names, numbers and internal citations were provided based on unverified online sources. The financial services sector is not immune from these types of incidents if generative AI tools are not used appropriately with a clear understanding of the source data. AI tools built on poorly managed inaccurate or incomplete company data can also provide outputs that suffer from similar “hallucinations”.

The rapidly evolving nature of AI tools means that means that both the value and risks are unclear to many users. Firms that do not properly articulate the value and limitations of AI may face inertia amongst workers. It is important to demystify the technology and show how it can improve work experience whilst setting out a framework for appropriate usage that aligns to client and regulatory expectations.

Training and upskilling workers can help explain the fundamentals of AI and teach hands-on skills in using these tools within their job functions, bridging any existing skills and knowledge gaps. This contextual understanding can showcase operational use of AI to assist with dull, repetitive tasks, thus opening up time for teams to focus on growth work that they enjoy and also adds value to firms’ progression.

Managing High Traffic and Data with Cloud

A recent report by Microsoft identified significant bottlenecks that can disrupt AI momentum; a key factor being low levels of compute capacity and adoption of background technologies such as cloud. Despite all its benefits, AI within data infrastructure does require substantial computing power and storage, making on-premise solutions cost-prohibitive. 31% of business have yet to adopt cloud and with the UK lagging other countries in digital technology infrastructure, businesses will need to become more familiar with this technology. Here, cloud computing emerges as a game-changer to grant businesses the flexibility needed to keep sensitive data secure while providing the computational power needed.

Leveraging cloud-based data management tools allows firms to store, process, and scale to accommodate increased traffic, which is particularly beneficial for handling data loads during high transaction periods. This ensures smooth user experiences and utilises decentralised networks for distributed low costs.

The successful deployment of such cloud-based services can help financial companies process enormous amounts of customer data and connect them with AI-processing capabilities without investing in expensive servers. With its ability to accommodate various infrastructure and AI capabilities, cloud solutions can easily handle changes in data to deliver unparalleled employee and customer experiences.

Future Success with Strong AI Foundations

A well-structured and forward-thinking approach to AI is essential, but the quality of AI outputs will only be as good as the data infrastructure that supports them. With the right foundation, even the most advanced AI systems will be able to deliver actionable insights. While it is a complex undertaking, a supported data infrastructure can yield significant rewards for improved decision-making and enhanced customer experiences.

A holistic approach encompassing AI technology, processes, and people can build an AI-ready data infrastructure, allowing financial institutes to remain competitive and adapt to evolving demand.  This will secure their position in an increasingly competitive market and ensure sustained success as the financial industry continues its digital transformation.

Continue Reading

Business

Technology’s Role in Transforming Insurance: From AI to Cyber Risk 

Source: Finance Derivative

Authored by Samiul Chowdhury, Principal Actuarial Consultant, RNA Analytics 

The insurance industry is undergoing a significant transformation, driven by rapid advancements in technology. From property and casualty to life insurance, the role of digital solutions has never been more important. Today, it’s almost impossible to imagine a successful, compliant insurance business without technology at its core. 

But how exactly is technology reshaping the insurance landscape? And what does it mean for the future of actuarial work, AI, and cyber risk? Let’s explore. 

The Essential Role of Technology in Modern Insurance 

Technology is the cornerstone of the successful modern insurance business – whether property, casualty or life. It’s no longer optional—it’s essential! Operating a successful and compliant insurance company today without the help of software solutions would be a real challenge. Whether it’s managing customer data, meeting regulatory demands, or assessing risk, technology is at the heart of everything modern insurers do.  

In recent years, regulatory compliance has been a top priority for (re)insurers across the globe, with IFRS 17 probably the number one focus. The new accounting standards are highly complex, and their implementation has forced many insurers to rethink and redesign their entire approach to financial reporting and infrastructure. However, this challenge has also been a catalyst for technological innovation.  

One of the most significant changes brought about by IFRS 17 is the integration of traditionally siloed such as functions such as actuarial, finance and accounting functions. This alignment gives insurers unprecedented insight into opportunities and risks, enabling them to make more informed decisions. Beyond compliance, accuracy and extensive flexibility, this integration offers insurers a chance to enhance accuracy, achieve greater flexibility, and gain a deeper understanding of their financial landscape. 

How AI is Changing the Actuarial World 

Much has been said aboutArtificial Intelligence (AI) and its potential to disrupt industries. In insurance, AI is already proving to be a game-changer, especially in actuarial work. With the right approach, AI holds great promise of making processes smoother and bringing faster, more accurate decision-making into play. 

However, AI is not here to replace actuaries. Instead, it enhances actuaries’ roles by automating their routine tasks such as data pre-processing, model fitting, and report generation. This automation allows actuaries to focus on more strategic tasks, giving them a more central role within the organizations. 

Meanwhile, AI modelling introduces new sources of uncertainty. Actuaries must understand the limitations and assumptions behind the AI models they are using. It’s important to ensure that these are fair, unbiased, and ethical —particularly when it comes to pricing and underwriting. This means actuaries will need to pick up new skills, especially in data science and programming languages like Python and R.  

In other words, AI offers actuaries the chance to work more efficiently and strategically, but only if they are prepared to navigate the complexities it brings. 

The Growing Challenge of Cyber Risk. How Do Insurers Keep Up?  

Cyber risk has emerged as one of the most significant threats insurers face today. Cyber insurance is not the same as it was twenty years ago. The policies were relatively simpler, and insurers didn’t have as much data or experience to rely on. Today, they are more complex, reflecting the increased scale and sophistication of cyber threats. 

As cyberattacks have increased, so has our ability to model and understand them. Insurers have gained more data over time, which has allowed them to get a better grip on the risks involved. However, here is the thing: technology evolves, and so do the threats. Whether it’s a data breach, ransomware attack, or even non-malicious technical failures like the recent CrowdStrike outage, the risks are more systemic and far-reaching than ever.  

Looking ahead, as we enter the Web3 era where information becomes ever more interconnected and managed by semantic metadata, we’ll have a complete set of new vulnerabilities. Business models will shift, and with that, the risks insurers will need to cover. By 2044, cyber insurance policies will probably look quite different from what we see today. 

Conclusion 

The insurance industry is at a turning point, driven by the rapid adoption of technology and the increasing complexity of risks like cyber threats. To stay ahead of the curve, insurers need to embrace AI, data-driven decision-making processes, and advanced risk models. 

Continue Reading

Copyright © 2021 Futures Parity.