The noise around generative AI (GenAI) is almost deafening with daily announcements of the commercial impact GenAI is having and of new capabilities, tools, and initiatives. To help make sense of ‘predictive text on steroids’, we asked futurist and long time friend of the FSF Rohit Talwar on how to advance the use of GenAI in financial services marketing.
What are current and emerging marketing leaders saying?
Many marketing leaders expect AI to have a significant impact. However, they also want clarity on how to use it, potential benefits, getting started, and addressing risk management and compliance concerns. These views were born out last month at FSF’s emerging marketing leaders’ Next Generation Summit.
Although most had little or no GenAI experience, they surprised themselves with what they created in a one hour exercise without any training. Participant teams received a one line challenge and then used different GenAI tools to turn these into marketing briefs, market propositions, advertising text, promotional videos, visuals, and radio ads. This led to healthy discussions about the how to enable successful adoption of GenAI.
Is GenAI having a commercial impact?
While end user organisations are reporting high levels of interest and experimentation, the resulting commercial gains aren’t yet being communicated. On the supply side the impact has been dramatic. Microsoft’s market capitalisation has risen almost $1.5 trillion (180%) since the start of 2023 following it’s $10 billion investment in OpenAI – the creator of ChatGPT. Even more spectacular is the $2.7 trillion (840%) increase in market capitalisation over the same period for Nvidia – whose AI hardware powers GenAI. On the services side, Accenture recently reported GenAI revenues of $1.1 billion in the first half of its fiscal year. The scale of investment and speed of product development suggest that we are very early in the commercialisation of AI – but that shouldn’t stop us from engaging with and learning about what GenAI can do for us personally and professionally.
What benefits could GenAI bring for financial services marketing?
At the core, GenAI is allowing firms to combine external and internal data to drive major gains in task efficiency, creativity, turnaround times, quality, and overall productivity. One of the biggest benefits is the speed with which marketeers can produce the first draft of marketing briefs, plans, text, and visual assets – cutting turnaround times by 90% or more.
Current applications include research, competitor and sentiment analysis, customer segmentation, performance monitoring, predictive analytics, ad campaign optimisation, dynamic pricing, and tailored offers. At the execution level, GenAI is enabling highly personalised and multilingual text, audio and visual content creation, content generation for SEO, A/B testing and experimentation, and chatbots and virtual assistants for customer support. Common internal applications include planning, training, event design, and meeting management.
Critical here is to treat GenAI like a smart, fast learning, and infinitely tolerant intern – guide, stretch, encourage, refine, and always validate the outputs.
How can we use GenAI to maximise the value of data while addressing security and bias concerns?
Benefits – Our data offers some of the biggest GenAI opportunities and risks. A major issue with many analytics project is that we don’t have clean and comprehensive data sets to support statistical analysis and more sophisticated AI applications. GenAI can help through data augmentation – generating synthetic data to mimic the statistical properties of real datasets. The tools can fill in any gaps with plausible data based on the patterns and correlations observed in the existing dataset. This approach can also help train AI / machine learning models to recognise what constitutes normal behavior and directly detect anomalies by comparing real data to generated data. GenAI can also consolidate and analyse multiple data sets to generate comprehensive reports, including visualisations, animations, summaries, narratives, insights, key findings, and potential implications.
Risks – There is a major concern over GenAI models putting sensitive company data into the public domain. The use of simulated datasets can overcome some concerns over entering ‘real’ customer or internal data. The key is evaluating different tools to identify which can be used or adapted to ensure that the data is retained within the organisation’s ‘walled garden’ and meet all existing data security requirements. Key here is model explainability to ensure that GenAI models are interpretable and transparent.
Bias – Research suggests that there are potentially 175 different forms of conscious and unconscious cognitive biases – making it almost impossible for humans to ensure total compliance. Hence, as GenAI models draw from the internet and our internal content, there is the risk of replicating existing biases in the outputs they generate. On the positive side, GenAI can be used to help address these concerns by asking the tools to evaluate content against a comprehensive set of biases to highlight potential issues.
Engaging risk and compliance
Many of the Next Gen group said their organisations had blanket bans on the use of GenAI. The key way in which a more nuanced approach can be adopted is through training. The first stage is to create common offsite learning experiences for colleagues in risk, compliance, and wider business functions to show them the capabilities of these tools. Critical here is using computers that are not connected to the company network! A common understanding of how GenAI works and what the tools can do allows for informed discussion of how they can be used, how to address genuine concerns, and how to involve these functions throughout the process of tool specification, evaluation, selection, and adoption.
So how can marketing departments get started with GenAI?
Training is key here, with an emphasis on the team learning together, and continuing to advance their skills and understanding of GenAI as new capabilities and tools emerge. Key here is choosing a few initial well defined tasks with which to get going. These need to be ‘safe’ activities that require no internal data – such as external research, competitor analysis, developing plans, designing training programmes, and planning events.
We are in the very early stages or AI development and adoption and the pace of both is likely to accelerate dramatically in the next few years – so we cannot ‘wait until things have stabilised’ or hope that regulation will protect us. Whatever the internal barriers to adoption, the key right now is understanding the capabilities of these tools and learn for ourselves how they can be used across our core tasks. Armed with that learning, we can then start to have informed internal discussions about how to get maximum possible benefit with zero associated risk.
The FSF has partnered with Rohit to deliver an introductory hands on training day for marketing executives on using GenAI to drive productivity and performance. You can find further details on the programme here.