Boston Consulting Group experts Nihmal Marrie, Caio Anteghini and Isme Oosthuizen stress the transformative potential adopting generative AI can have on finance functions.
The financial landscape, much like other sectors, is poised for a revolutionary makeover through the power of generative AI tools. While the potential is dazzling, finance leaders need to navigate beyond the hype and gain a deep understanding of how these tools will redefine the future of financial operations.
Exploring the generative AI journey in finance
According to Boston Consulting Group (BCG) MD and partner Nihmal Marrie, the trajectory of adopting generative AI in finance resembles the pattern of an S-curve (see Exhibit 1 below).
“Currently, finance teams are grappling with how this technology can enhance existing processes, particularly in terms of text generation and research. Yet, a more profound transformation lies ahead – one that will reshape core processes, rejuvenate collaboration strategies, and effectively address risk,” he says.
He explains that a vision is emerging where generative AI collaborates seamlessly with traditional AI forecasting systems, producing insightful reports, explaining variances, and offering strategic recommendations. “This amalgamation promises to elevate the finance function’s proficiency in forecasting and strategic decision-making, ultimately boosting operational efficiency.”
Navigating the challenges in generative AI adoption
While the prospects are promising, integrating generative AI into finance functions is not without its challenges. Accurate data analysis and security concerns loom large.
According to BCG partner Caio Anteghini, to pave the way for seamless adoption and stay ahead of the curve, CFOs must acquaint themselves with impactful applications of generative AI in finance and lay the groundwork to harness its evolving capabilities.
“Currently, generative AI tools excel in processing text and images. However, their capacity to generate accurate numerical analyses – a cornerstone of finance – is still evolving,” he explains.
“These tools are capable of initiating analyses on limited datasets, but they require refinement. In contrast, traditional AI in finance adeptly handles numerical data for forecasting and risk assessment. This gives rise to a dual landscape where some applications are tailor-made for generative AI or traditional AI techniques, while others present opportunities for harmonious integration.”
Caio says that, presently, the integration of generative AI into finance functions focuses on augmenting existing processes. This includes tasks like narrative generation and one-off analyses of small datasets. Current and upcoming applications across the finance spectrum encompass:
- Finance operations: Generating preliminary drafts for text-heavy tasks like contract drafting and supplementing credit reviews.
- Accounting and financial reporting: Offering initial insights during financial statement iterations at month-end closures and aiding audit trails.
- Finance planning and performance management: Conducting ad-hoc variance analysis by comparing actuals to plans and producing reports to elucidate unit financial performance for business partners.
- Investor relations: Providing comprehensive support for various aspects of quarterly earnings calls.
Overcoming the adoption hurdles
Relative to preceding technologies like robotic process automation and process mining, the barriers to experimenting with generative AI in finance are comparatively low, says Ciao. Nevertheless, addressing specific challenges is essential to fully harness the technology’s potential. These challenges include:
- Data accuracy: Early iterations of generative AI tools can struggle with precise calculations. Ensuring accuracy demands meticulous design, or alternatively, using workarounds to generate content based on calculations performed outside generative AI tools. Advancements like GPT-4’s improved capabilities underscore the potential for progress.
- Data security: The use of public cloud platforms for training generative AI models poses the risk of leaking proprietary data through security breaches.
- Governance model: Generative AI tools lack context awareness and real-time information. Developing an explicit governance model for output validation is currently lacking.
- Hallucinations: Generative AI occasionally produces convincing yet incorrect responses.
The transformative potential of tomorrow’s Generative AI capabilities
As generative AI evolves to accurately analyse extensive datasets and finance professionals become more adept at utilising this technology, a new generation of AI-driven ‘copilots’ or ‘assistants’ will emerge.
“Traditional AI and generative AI will coalesce to create synergistic use cases. For instance, while a traditional AI forecasting tool generates projected financials, generative AI can explain discrepancies and, crucially, offer recommendations for diverse forecast scenarios and corresponding business decisions,” says BCG associate director Isme Oosthuizen.
Consequently, she explains, the next phase of finance co-pilots will empower the future finance function in three pivotal ways:
- Revolutionising core processes: A widening array of generative AI assistants will continually reshape core financial processes such as contract drafting, invoice processing, and general ledger reviews. These focused assistants will initially enhance the efficiency of specific processes by approximately 10 to 20 percent. As capabilities evolve, they will encompass a more substantial portion of overall finance tasks. Over time, generative AI will seamlessly integrate with currently manual or cumbersome processes.
- Reshaping business collaboration: Generative AI will provide indispensable support to finance's business partners. This support extends to financial forecasts, scenario planning, and streamlined business intelligence. Arduous finance activities that impede insight extraction will be revamped, enabling swift and clear insight generation. Pairing generative AI with traditional AI will amplify these capabilities.
- Risk management and mitigation: Finance teams already leverage AI in audit and control environments to identify anomalies indicative of fraud or noncompliance. The next wave of generative AI could predict and explain anomalies, allowing timely identification and communication of associated risks to avert undesirable audit findings.
CFOs: Agents of preparation
CFOs should be proactive and embrace generative AI as a critical step toward maintaining a progressive finance organisation. Isme says CFOs can adopt a proactive stance by:
- Creating proofs of concept: Experimenting with low-entry use cases, such as investor relations and contract drafting, refining approaches iteratively.
- Training internal talent: Identifying skill gaps, providing training, and recruiting individuals equipped to handle evolving use cases.
- Developing AI capabilities: Cultivating AI competencies in-house by establishing centres of excellence or integrating AI skills within technology teams.
- Collaborating with IT: For successful implementation, forging a robust partnership with IT is imperative, addressing security concerns and prioritising AI investments.
- Championing Generative AI: Collaborating with peers to identify cost-efficient use cases beyond finance, allocating investments to generative AI, and integrating AI-influenced cost targets into business plans.
“Generative AI’s rapid evolution demands immediate action,” adds Nihmal. “Excelling in text generation and rapidly advancing in numerical analysis, finance leaders must proactively monitor AI’s development, gain hands-on experience, and foster organisational capabilities. Given the accessible entry barriers, there is no need to wait for further advancements.”
CFOs should wholeheartedly embrace this technology, surmount adoption obstacles, and inspire their teams to harness generative AI’s potential across the finance function.