Generative AI and its talent and change implications within financial services - 1 of 3: what is generative AI and what is its impact on work?

Introduction

My next few posts focus on generative AI and the impact on talent and change within financial services.

There are moments where changes build up slowly and then they suddenly erupt into the public consciousness and change the world. Think about the world of streaming before Netflix and Spotify (I am still mourning my CD collection). Generative AI is in one of those moments.

Banking, insurance and other financial services sectors are no strangers to technology change, from being the early adopters of computing and the first ATMs, through to more recent waves of robotics, AI and real time platforms. Generative AI has the potential to drive as much change for the industry, if not more so, than what has gone on before.

Generative AI has the potential to transform business and elevate human potential at work, if applied responsibly and ethically. The banking and insurance sectors have the highest potential for gains from Generative AI – and some of the greatest risks too. So it’s not just a ‘tech trend’ for the CIO and CDO – this is a ‘business agenda’ item that needs CEOs, business leaders, risk, HR and change professionals fully at the table and engaged in shaping the change ahead.

This latter group is who this blog is for. This topic is moving fast, but hopefully this is a helpful impetus into the decision making and experimentation in your organisation.

In this first blog, I focus on what Generative AI is and its overall potential impact on work and talent in financial services. The next couple of blogs focus on the specific applications in financial services work and how we can approach this change well.

What is Generative AI?

Artificial intelligence isn’t just one thing nor is it new. It is used as a catch all term for a range of technologies working together to enable machines to sense, comprehend, act, and learn with certain human-like levels of intelligence (or in some cases better).

AI matters because it’s an incredible source of business insight, agility and efficiency in business. And it’s accelerating quickly, powered by advancement in deep learning and neural network architectures, massively increased computing power and interdisciplinary approaches. This is why 84% of business executives believe they need to use AI to achieve their growth objectives. Some of you will be experts already; some of you will be exploring AI for the first time – there are some useful resources here to find out more.

Generative AI has made recent leaps forward and caught the public and media’s attention. Awareness of generative AI is growing rapidly, popularised by ChatGPT from OpenAI, which took 5 days to reach one million users and two months to reach one hundred million active users. Microsoft followed with their Co-pilot announcement for Windows, Google with Bard and so on.

But what does generative mean? ‘Generative’ refers to the AI learns how to ‘create’ new data, not just recognise it. The data created might be written text or code, images, speech, music and video, 3D or gaming content (or a ‘multi-modal’ mix of these). The models synthetically generate new content that are based on the original artefact or data set, but do not repeat it, creating results that are varied in style and substance. Using probabilistic models the algorithms learn complex patterns and structures in training data and use that to generate new data similar in style, content and context. They can do this in seconds and get excitingly/alarmingly close to the quality of human-generated outcomes.

To do this, generative AI is based on a foundation model. A foundation model is pre-trained on vast amounts of data. In particular, large language models have jumped forwards in maturity, now being able to learn language, context and intent and be independently generative. These models can read, comprehend and decide based on input data and user questions, including multiple steps within a ‘thought process’ and ‘learning’ from past processes.

Once established these foundation models can be fine-tuned and adapted for a specific usage through training on much smaller amounts of data. This starts to overcome many of the previous challenges for organisations who lacked the data volume and diversity to train AI well.

What is the impact of generative AI on work?

Generative AI has the potential to perform various types of human work, in areas of intelligence like vision, language and reasoning:

This matters for talent professionals because language tasks make up 62% of the time employees work, of that, 65% can be transformed into more productive activity through augmentation and automation using generative AI.

While it has more ‘human like’ capabilities, to use generative AI effectively you still need human involvement in the process. Often the generative AI is the augmentation ‘copilot’ to the human colleague, providing draft answers or inputs for the human to use. As the AI will sometimes get things wrong, when it automates human work or decision making you still need a ‘human in the loop’ to check the results.

With every advancement in technology there are bleak predictions regarding jobs displacement; mixed with the hope of new industries, growth and job creation. A recent Goldman Sachs report predicted 300 million jobs would be exposed to automation by generative AI: putting some at risk, but also raising productivity in existing work and generating new jobs, potentially increasing global GDP by 7%. The truth is that generative AI is neither inherently good or bad for work. Instead the consequences will be down to how business applies generative AI to existing work and new opportunities – and some of the internal and regulatory guardrails to this. This area is evolving rapidly: no one has all the answers and there will be unintended consequences along the way. So using it well means experimentation, iteration, reflection and learning.

One certainty is that to apply generative AI well it is not a ‘race to the bottom’ seeking to eliminate humans from all processes and maximise automation at all costs. Instead to drive human, social and economic value, we need to consider how generative AI may create new opportunities and elevate human potential, in the right work and where it adds value.

To do this we need to look more carefully at the tasks and activities that make up work. Only by doing this, can we really understand the skill needed for each task and the value of a human or machine doing it – or in many cases them doing the task together. We have honed these sorts of ‘human and machine’ analytics and strategies during earlier waves of AI and technology change.

Paul and the other authors’ recent blog on Harvard Business Review called ?‘Generative AI Will Enhance — not Erase — Customer Service Jobs’ introduces a useful framework to explore how we apply generative AI to work. They highlight that human, automated, augmented and emergent tasks are the ingredients around which companies should redesign jobs to get the maximum advantage from generative AI. They studied an existing customer service role and found four tasks that only a human could do, four that could be entirely automated by generative AI, four human tasks that could be augmented with generative AI to make the work more effective and five new high-value tasks emerging from the use of generative AI. In one simple job, all four impacts could be found and the full potential of generative AI is only addressable by analysing and reinventing work and skills, not just addressing the whole job.

Equally important is how we design meaningful human work and think about how we develop smaller but more highly skilled workforces. Previous waves of AI have focused more on automating repeatable tasks, leaving tasks that need more human creativity and imagination, problem solving, judgement, strategic planning and emotional intelligence. Generative AI is more human in its abilities, starting to address some of these; but it is still not human.

So we need to think even more carefully about where and when we want human work: where does this generate business and customer value – and how do we make this good work for people? Customers and clients value human connection at key moments, so thinking about when we consciously design human service into the process is essential. We already have an issue with job quality and meaningful work and growth are a key part of that. On our six ‘net better off’ indicators of human care at work, banking and insurance have gone backwards over the last three years.

This is where CHROs need to be engaged, working with their CIO and CDO colleagues and engaging business leaders in a meaningful way about changing work and skills:

“Leaders must begin now to do the hard work of reinventing jobs and creating the most effective mix of human, automated, augmented, and emergent tasks in the context of the company’s specific business.” Daugherty, Wilson and Narain, HBR, 30 March 2023

What is the potential impact of generative AI on work in Financial Services?

Based on labour market data we have done a preliminary analysis of what work can be addressed by generative AI. Most work in financial services is language-rich or data-heavy information, service and knowledge work, so it is not suprising that banking and insurance are high up the list of impacted industries:

Generative AI has potential to impact 90% of work time in banking and 88% in insurance (vs 62% cross-industry average).

Work with higher potential for automation makes up 54% of work time in banking and 48% in insurance (vs 31% cross industry).

Work with higher potential for augmentation makes up 12% of work time in banking and 14% in insurance (vs 9% cross-industry), where the generative AI models would need greater human involvement.

It’s important to note this is potentially addressable time – for benefits to be realised the task needs to be found and aggregated - and in many cases only part of the task can be automated or augmented. Nonetheless, the impacts on work in financial services are considerable.

In my next blog I will start to look at some of the potential applications of generative AI to specific types of work within banks and insurers.

Summary

Generative AI, while hyped, is a leap forwards. It is already useful today and its capabilities are accelerating quickly. Its ability to generate content and be trained into specific domains, means it has significant application to the language and data work inside financial services. To realise the potential of generative AI and design good work for people, we need to look at the tasks that make up current jobs and decide where we want human work, where generative AI can do the job better and, most importantly, where human and machine can work best together. We also need to find the opportunities for new client propositions and experiences now possible using generative AI – and go through the same intentional design of future work for growth. This needs the CIO, CDO, CHRO and CEO to work together to seize the opportunity and move forward quickly and responsibly, learning as we experiment and scale up.

If you want to find out more about Accenture’s perspective on generative AI, please see our Technology Vision or our latest point of view on Generative AI. We’re working internally and with clients to apply generative AI to work, so please get in touch if you’d like to discuss how we may be able to support you.

This content is provided for general information purposes and is not intended to be used in place of consultation with our professional advisors. This document may refer to marks owned by third parties. All such third-party marks are the property of their respective owners. No sponsorship, endorsement or approval of this content by the owners of such marks is intended, expressed or implied.