Sanjay Gupta: Can Science and God Coexist?
Faith and science may often seem at odds with one another, but renowned geneticist and former NIH director, Dr. Francis Collins, says that he sees…
Thought Leader: Sanjay Gupta
AI In Financial Services: Only Users Will Survive
Artificial intelligence (AI) has long captured the human imagination. From Arthur Clarke’s seminal novel 2001: A Space Odyssey, to Arnold Schwarzenegger’s famous role as the Terminator, the idea of smart machines capable of learning and matching human intelligence fascinates us. With GPT-4 and other LLMs (large language models) catapulting the sector back to the top of news feeds, now is an opportune time to assess the AI financial services landscape. As with any breakthrough technology, excitement and media hype lend themselves to a healthy dose of mania. As a reminder, whether as investors or businesses, it is essential to remain clear-eyed about the potential impacts of AI on the industry sectors you care most about. The goal of this series on AI is to provide some context and understanding about the technology to ensure you do not get lost in the complexity.
Artificial intelligence (AI) has transformed numerous industries, with the financial services sector emerging as a prime beneficiary. In this article, we will explore the current state of AI in the financial services industry, key areas of application, recent venture capital investments, and the potential challenges and opportunities that lie ahead. We will first cover some of the key nuances shaping the AI conversation. The futuristic machines of our dreams remain a remote prospect, but major advances in LLMs are on the verge of upending all new and existing businesses, education and government
Over the next five years, additional breakthroughs in AI will reshape business processes and opportunities in the financial services sector. Investors see the potential for novel AI systems to address difficult industry challenges. This includes accurately evaluating asset prices, enhancing regulatory compliance and fraud detection, and automating insurance claim processing. As the market penetration of AI products and services increases, the gap between firms that successfully harness AI power and those that do not will come into sharp relief.
Let’s dive in.
Briefly: where AI technology stands today
In today’s rapidly advancing business landscape, it’s not uncommon to find oneself unsure of what exactly falls under the umbrella of AI. In fact, many individuals and organizations alike may find themselves grappling with this question. Rest assured, you’re not alone in this.
This is quite understandable as the world of AI can be quite daunting for those unfamiliar with the intricacies of the rapidly advancing field. The term “AI” encompasses a wide range of computing capabilities that attempt to replicate various aspects of human intelligence, such as reasoning, generalization, and learning from experience.
GPT-4 is a natural language processor and is part of a growing group of large language models (LLMs). By prompting the AI with text, the technology will scour all of the internet, online books, research or anything digitally available to generate responses. These responses can be factual, coherent and wide in scope. They can also be inaccurate. Yet, AI can be an amazing tool to enhance productivity and creativity.
Some AI evangelists envision a day when machines can operate autonomously, set their own goals, and control their environment. Achieving so-called “general artificial intelligence” of that nature may represent the apex of computing technology development, but it remains decades – if not centuries – away. Some within the computing industry question if we can ever get there at all. However, as the technology progresses, we’ll have AI models creating AI models that we humans may or may not fully understand what the outcomes the models are trying to achieve. It could be to end cancer or it could be to end the human race. A recent survey of coders working on AI expressed a 10% probability the technology would end the human race.
In contrast to general artificial intelligence, we are amid major advances in “narrow AI” – sometimes referred to as reactive machines. Narrow AI computer models specialize in solving a particular problem or providing a discrete service. It is a type of AI that is focused on performing specific functions, rather than trying to replicate the general intelligence of a human being. These models cannot connect disparate domains or knowledge or engage in complex abstract reasoning, but their development portends major shifts in capital and investment flows.
Narrow AI can be broken down according to its functionality. For example, natural language processing enables computer programs to analyze spoken or written words. The frustrating customer service voice robot that refuses to connect you to a real human is using natural language processing to understand your words. Another type of AI, computer vision, enables the capture, processing, and analysis of real-world objects. Without computer vision, your new Tesla would not be able to sense the pedestrians, bikes, and vehicles around it. Finally, generative AI is yet another capability. Generative AI, which powers GPT-4, can produce original media content by repurposing existing data sources. This tool can effectively and hyper-efficiently speed our ability to process information and enhance our decision-making skills. It can also be a boost to creativity whether it’s OpenArt for generating pictures or Boomy for creating music. Many of these AI websites are simply using GPT-4 or Dall-E as their LLM. It’s no small wonder that GPT-4 shuts down on Mondays due to traffic.
Another area of AI development is Explainable AI (XAI), which aims to create AI systems that can explain their reasoning and decision-making processes to humans. This is essential for building trust in AI and ensuring that its decisions are transparent and accountable.
What all these AI capabilities have in common is the ability to train themselves. Computers are getting increasingly skilled at completing complex, albeit narrow, tasks presented to them. As the models operate and receive feedback from the outside world, their performance improves. (The more specific and detailed the prompt, the better the user outcome.) This capability, known as machine learning, is what allows your iPhone to better anticipate whether you want to call your best friend at the office or on their mobile. Overall, AI technology has come a long way.
A new computing platform, powered by Big Data & Chips
Understanding what advances AI can realistically achieve in the next 5-10 years is one half of the equation. The other half is understanding the basic drivers behind AI technology developments. Taken together, various AI capabilities represent a transformational shift in the digital economy. The convergence of narrow AI capabilities will deliver more powerful, insightful computer programs that can vastly enhance productivity. That is why Microsoft CEO Satya Nadella refers to the disparate suite of AI technologies as the “new computing platform.” BTW, Microsoft invested $10 billion in OpenAi in January of 2023 and will soon be integrating it into its suite of products.
The emergence and growth of the AI computing platform have been facilitated by three key enabling factors: the availability of increased data volumes and the decreasing cost of computing via cloud computing and specialized semiconductor chips. AI models require vast amounts of data to learn from and sift through, and these factors have made this possible. The internet, mobile phones, and connected devices are generating enormous amounts of data that were simply not possible to collect in previous eras of computing. As opportunities for data generation increase, the cost of computing power continues to fall. That is good news for AI software developers, since AI models require vastly more power than older technologies to function effectively. Without advancements in hardware and software that make computing power more affordable, running complex AI models would be cost-prohibitive for many firms.
Moving forward, the companies that successfully harness AI capabilities will accelerate their success via enhanced creativity and better decision making. AI will help these firms increase productivity, generate powerful customer insights, and improve risk management. These advantages will help firms with AI competencies outpace digital laggards who fail to adapt. Shortly, there will be two types of businesses, those that use AI and those that have gone out of business due to not using AI. Remember how Netflix’s adept integration of new video and streaming technology dealt a death blow to Blockbuster’s conventional business model?
The good news for investors is that as AI technology continues to advance and gain mainstream acceptance, there will be a plethora of opportunities for investors to cash in on the broader AI trends. Some of the primary areas where investors can look to make a profit include AI-powered companies, technology infrastructure providers, and professional services firms that specialize in AI development and implementation. Investing in established companies that are incorporating AI into their existing business models is one promising route. Indeed, the heaviest spenders on AI technology include tech giants like Amazon, Alphabet, Meta, Alibaba, and Baidu. It is difficult to accurately estimate the size of internal investments, but company announcements indicate that significant portions of research and development dollars are being devoted to building AI competencies.
Entirely new businesses will also spring out of the advances in AI and computing. CB Insights breaks these upstarts into three categories: AI development tools, industry-specific applications, and cross-industry applications. All three segments will see the rise of new companies that leverage AI capabilities in exciting ways. AI software developers will help businesses apply expanded analytical power across all business operations, while companies will leverage AI to build compelling new products in their industries.
Finally, companies that help supply the infrastructure required for powerful computing are poised to profit handsomely from the increased demand for AI products and services. This includes cloud service providers like Amazon Web Services and Microsoft, as well as hardware and chip manufacturers like Nvidia and Taiwan Semiconductor Manufacturing (TSM). TSM is already garnering between 15-20% of revenue from AI-related demand, and the mix is likely to be higher in the future.
Impact on the financial services sector
Artificial Intelligence (AI) has already made a significant impact on the financial services sector and is expected to transform the industry even further in the coming years. AI has the potential to revolutionize many areas of the financial services sector, from fraud detection and compliance to investment decision-making and customer service.
To better understand how AI advancements are likely to play out, we now turn to VC investment data in the financial services sector. Bankers, asset managers, and insurers are already integrating AI capabilities into their services, particularly with customer service. Virtual assistants and chatbots were the first frontiers, but AI power is starting to rapidly go into other operational functions. Enhanced operational efficiency will be good news for companies and consumers alike. The industry is expected to save more than $1 trillion by 2030 thanks to AI, with traditional financial institutions shaving 22 percent from their costs. But there is more to the story. Narrow AI advancements offer particularly promising solutions to tough financial services challenges – and the data provides clear clues on the areas of the highest impact.
Better asset management through improved models and automated functions
A central challenge for asset managers and market analysts is the integrity of their models. Whether you are a more traditional securities broker or a developer of a high-frequency trading algorithm, accurate forecasts of market movements are essential. Yet most modeling capabilities are stuck using linear regression models that do not adequately account for real-world relationships. At the same time, more data than ever before is spewing out of digital news outlets and social media websites. The ability of conventional models to capture and process reams of unstructured data has not kept pace.
AI modeling capabilities aim to change the status quo. They can help improve asset management through advanced models and automated functions. With the help of AI-powered tools and algorithms, financial institutions can analyze vast amounts of data and generate insights that can be used to make more informed investment decisions.
Four companies that deserve special mention hauled in over $65 million in venture funding to create new analytical investment models that can better account for non-linear relationships (read: real-world complexity). They are Boosted.AI, SESAMm, MioTech and Cervest.
Boosted.AI and SESAMm are both using AI to harness new data sources and improve the prediction of market developments. They hope to help fund manager clients create more value in their portfolios and improve investment decision-making. In addition to improving model integrity, these AI investment platforms also enable greater stress testing of scenarios. For these firms, increases in computing power translate to an ability to model more possibilities simultaneously.
On the other hand, Hong Kong’s MioTech and London’s Cervest are aiming for a narrower slice of the investment modeling market. They are using their AI capabilities to help investors make sense of climate risks and environmental, social, and governance (ESG) developments. Climate and ESG dynamics are increasingly part of the decision-making equation for investors. However, accurate assessments of how these risks impact a firm’s outlook are especially challenging. On this front, the AI models from MioTech and Cervest help draw out insights from large volumes of data sources, including videos, annual publications, and industry reports. Natural language processing is a particularly important component of these AI models, which are helping uncover the critical links between climate, ESG, and business performance.
Powering new approaches to compliance, fraud detection, and risk management
Digital connectivity and rapidly evolving regulatory landscapes are increasing the risks for financial institutions. Yet according to a survey from consultancy Deloitte, only 4% of firms are using AI in their legal and compliance operations. VC investors view this gap as an untapped opportunity. That is why they funneled $200 million and $45 million to Feedzai and Unit 21 respectively. Both firms are harnessing the power of AI to help the world of finance adapt to new compliance realities. The firms primarily focus on preventing fraud and money laundering in transactions, but their models could be adapted to root out other internal compliance issues.
Traditional fraud detection programs are built using rigid rules and return a high number of false positives, wasting time and resources that could be used in investigations. AI’s ability to analyze large volumes of data quickly and accurately will have a massive impact in these critical areas, enabling financial institutions to better manage risk, reduce fraud, and comply with regulations. AI models promise to improve their functionality over time, freeing up more human brainpower for higher-value activities. Someday soon, generative AI capabilities may even be able to create the first drafts of legal briefs, regulatory filings, and other compliance documents. Humans would still be needed to set up the model and edit the output, but the gains in productivity would be immense.
In addition to fraud and compliance, AI models are being applied to improve risk management decisions that underpin lending. Zest AI pulled in a $50 million investment to accelerate the commercialization of its software model that improves credit underwriting. The company’s explicit goal is to build a more inclusive financial system by drawing on more data to help demonstrate creditworthiness. Legacy credit modeling approaches fail to account for the wide variety of data that could help underserved consumers obtain loans.
A similar dynamic exists for start-ups and small businesses that may be excluded from commercial lending programs at traditional banks. Zeni, a firm that aims to build an AI finance concierge for these underserved firms, attracted $34 million from VC investors. The emergence of AI models that can expand credit access will help consumers and small businesses who face daunting financial challenges. It will also force traditional banks to become more competitive or risk losing out on profitable lending opportunities to unconventional lenders.
AI support for insurance underwriting and claims processing
Enhanced risk modeling and process automation hold similar promises for the insurance industry. AI insurance start-ups are building models that can be trained to administer specific processes, transform data, and interact with other information technology stems. These models are particularly useful in improving underwriting, as they can analyze vast amounts of data to identify patterns and predict future outcomes. By automating many of the underwriting processes, insurers can reduce errors, increase accuracy, and ultimately make better decisions about which risks to accept and at what price. However, VC investment points to a greater market opportunity for claims management.
The evaluation and administration of claims is a major driver of workloads at insurers, regardless of whether they serve retail or commercial customers. Evolution IQ, which received a $21 million Series A investment last April, is building a product that can provide insurance claim guidance for companies. Likewise, Tractable is developing AI software that uses computer vision AI to assess damages from accidents and estimate repair costs. The London-based firm’s most recent investment round hauled in $60 million. The manual work required to review several insurance policies and analyze claim details before deciding can be automated by AI, and then reviewed by a claims agent for accuracy.
Wrapping it up
The emergence of narrow AI and LLMs has brought significant advancements in business operations, creating new markets for both established players and new entrants. In the financial services sector, the potential for improving decision-making in investment, lending, insurance claims management, and compliance has given rise to a host of opportunities. Access to high-quality data will be one of the biggest concerns for firms. Poor-quality data will negatively affect the performance of AI models. Thus, firms that can afford large investments are likely to expend significant capital building foundational data models to control large, proprietary datasets, making them indispensable players in the future digital economy. Meanwhile, nimble AI developers will identify profitable niches to provide innovative solutions to analytical problems.
As digital markets and technology adoption deepen globally, AI will continue to transform the business landscape, and its impact on markets will be profound. Bill Gates recently wrote, “Artificial intelligence is as revolutionary as mobile phones and the Internet.” The US and China are poised to be dominant nation-state actors in AI development, accounting for 80% of global VC investment in the sector. Strong competition in the AI marketplace and geopolitical arena will define how AI technology evolves. Although we are far from fully autonomous computers, narrow AI advancements are already transforming our world. Businesses need to brace themselves and be prepared to adapt to these changes to succeed in the future.
The emergence of narrow AI has brought significant advancements in business operations, creating new markets for both established players and new entrants. In the financial services sector, the possibilities for enhancing decision-making on investments, lending, insurance claims management, and compliance represent the most compelling emerging opportunities. However, as the market penetration of AI is still relatively new, strategic choices for firms remain.
One of the biggest questions is how firms will choose to secure access to high-quality data. “Garbage in, garbage out” will continue to be the adage of data scientists and AI developers. If you feed your model bad training data, you can expect poor performance outcomes.v
With quality data serving as the coin of the realm, firms that can afford large, up-front investments are likely to expend significant capital building foundational data models. Controlling large, proprietary datasets will make them indispensable players in the future digital economy. At the same time, nimble AI developers will see ample opportunity for profitable niches. The start-ups profiled here are mostly building on top of established datasets, yet they still hope to deliver innovative solutions to intractable analytical problems. A full coterie of professional services firms, advising on all aspects of AI capability development and implementation, will also evolve to support the burgeoning sector.
The impact of AI on the financial services sector has already been significant, and there is no doubt that AI technology will continue to transform the industry in the months/years ahead. The use of AI for enhanced risk modeling, process automation, compliance, fraud detection, and claims management is already improving decision-making and operational efficiency in the industry. Moreover, with the increasing availability of high-quality data and the development of new AI technologies, there are numerous emerging opportunities for incumbents and upstarts alike. As digital markets and technology adoption deepen in most corners of the world, the AI transformation will leave a profound impact on companies, markets and the economy.
And the US and China look set to be the dominant nation-state actors, with AI start-ups in both countries accounting for 80% of the global VC investment in the sector. Robust competition in the AI marketplace and the geopolitical arena will shape the contours of how AI technology develops. We may be a long way from fully autonomous computers, but advances in narrow AI/LLMs are now ushering in an exciting new world. Firms that can effectively harness the power of AI will be well-positioned to gain a competitive advantage in the marketplace and thrive.
Relevant and recent posts
Marc Short on what to expect from DOGE in relation to next year’s legislative agenda
Marc Short comments on what to expect from DOGE in relation to next year’s legislative agenda for CNBC. Looking for a great keynote or public…
Thought Leader: Marc Short
Marc Short on whether Musk has permanently changed how congressional communication functions
In this video, Marc Short joins CNBC to discuss whether Musk and Twitter have permanently changed how congressional communication functions and what to expect for…
Thought Leader: Marc Short
More with this speaker and recent posts
AI In Real Estate: The Future Of Property
Thought Leader: Andrew Busch
Sanjay Gupta: Can Science and God Coexist?
Thought Leader: Sanjay Gupta