Microsoft released last week a video entitled “The future of work with AI” showing how ChatGPT is being embedded in the Microsoft 365 series of apps. Inspired by it, I put together my thoughts about how AI is likely to change cognitive work, looking specifically at corporate banking to understand and anticipate what’s coming.
Wake-up call
A few years back, my boss, who was at the time head of Corporate & Investment Banking in Nordea, asked that we write an internal blog about AI and its impact on us, as a way to familiarise colleagues with tech trends and their impact on banking. AI was on a sort of hype and it felt like it was about to change everything. After a few generic articles on this topic, we however concluded that use cases were really few in our sector of activity. Not to offend my manager, we kept the name, the AI blog, but we discreetly pivoted the focus to more general technology developments and how we could make good use of them.
Four years have passed and boom, ChatGPT has woken up even the most sleepy technology followers. If you have tried it, I bet that you have been impressed too. And when you listen to Satya Nadella and his colleagues describe how they are embedding it in all their Microsoft 365 suite of apps, from Word, Excel, PowerPoint and Teams, you feel that it is no longer a question of hype. Something is really happening.
What’s new?
The main change is that AI has so far silently been working in the background, like in web searches and social media to optimise recommendations, or in banking operations by for instance powering the algorithms tracking money laundering.
But this new generation of AI is different since it is accessible through a very easy interface for us, natural language. As a result, it can be used more concretely to facilitate many tasks, from making a research on a certain topic, summarising a discussion on Teams, to creating a PowerPoint presentation from a Word document (and vice-versa), analyse complex data or draft a business strategy. And much more than that.
Microsoft calls it the Co-pilot, which is a clever branding. To me, it is closer to an AI personal assistant since in order to be efficient, it needs to have access to all your (work related) data, from emails, business chats, documents, calendars, contacts and CRM. And when it does, it can bring the context of your own data into what you are working on, putting together connections and insights in a way that was so far out of reach.
Satya Nadella claims that it will transform work as we know it. Probably, but at the minimum and for office workers, it looks like it will provide a leap in individual productivity growth.
I hear some blasé specialists claiming that we are not yet there, that ChatGPT is not that powerful, that it makes mistakes or provides sometimes misleading answers. In fact, Microsoft is the first to admit that when it is brought to their suite of apps, mistakes will be made there too. At this stage, it is a supporting tool and all the outputs need to be carefully checked and often complemented manually.
In any case, this debate is secondary for me. It is the trend and its acceleration which are impressive. ChatGPT is regularly fast publishing new versions, going further in the sophistication of what it can do. In addition, when it is brought to Microsoft 365 and its apps, the contextual data will be automatically taken into account, further lifting the quality of what it produces. Besides, the fierce competition by Google and others will ensure that developments in this field will continue on an exponential basis.
What’s in it for corporate banking employees?
Taking the analogy used by BCG in one of their Corporate Treasury Insights reports, the job of a corporate treasurer, the main counterpart of the corporate banking front line, can be split in two: the morning treasurer and the afternoon treasurer, with completely different expectations from their bank.
The morning Treasurer just wants an automated silence from their bank
The morning treasurer is working on everyday operations and wants their bank to “make their life easy”. They are aiming at the so-called “zero interaction model” and in fact just wish silence from the bank. This covers areas like automation where they expect, over time, full bank integration in their systems and value chain, painless processes like KYC or onboarding, transparency including automatic reporting, and a much higher degree of self service. This is where process automation, API integration and “intelligent robots” will increasingly take care of the tasks where in fact banks’ relationship managers don’t add much value. And it will free up their time.
Only AI augmented relationship managers will match the needs of the afternoon Treasurer
The afternoon treasurer is working on strategic issues and wants their corporate bank to “make them smart”. This is where the role of the relationship manager is emphasised as long as they can become their “trusted business partner”. Customers will expect market intelligence and advisory to a greater extent than today, with proactive, relevant and customised communication from their bank. This is feasible only by making a much better use of data, the one banks have and the one they need to get, and using intelligent tools that provide insights specifically relevant to them.
From tailor made to generic AI tools
Some banks have been working to develop those intelligent tools, capable for instance to go through all the quarterly and annual reports published by a corporate customer and extract the relevant information. Or to provide a news feed scanning all the information available on the internet and picking up the relevant ones such that it could for instance deliver every morning the three most relevant news items the relationship manager needs to know about each of her customers before calling them. Or to build a control dashboard with a summary of all the interactions with a customer, including the status of internal processes and the expected time for resolution. Or to make available an alert and lead functionality with a Machine Learning based feed-back loop to ensure that it would be increasingly relevant to the relationship manager.
Those are just examples of what is already there or coming. But an important point in my opinion is that those tools won’t need to be built individually by each corporation as has been the case until now. They can soon be procured as generic tools and then adapted to the bank’s specific needs. I am curious for instance to see what Microsoft Dynamics CRM will be able to do once it is powered by ChatGPT AI and connected to all the relevant corporation’s internal and external data. It will for instance definitely kill the self made and incredibly clumsy customer chatbots still used by retail companies today, since once a chatbot like ChatGPT has access to the company’s context, it can do a much better job.
And from the moment generic tools can be applied, development time and costs decrease dramatically, further accelerating the use of AI in daily working life.
The customer data confidentiality issue
Some banks have banned the use of ChatGPT by their employees, worrying that if they insert customer sensitive information in the chatbot, they could be in breach of the very important customer data confidentiality issue. I understand the concern, but it is a bit the same discussion as with cloud computing a few years back when banks were initially very resistant to outsource their server capacity, for primarily the same reason. To me, it is only a question of time before this issue is resolved. Besides, I am curious to see how banks can continue to ban it once it is embedded into employee tools like Word, Excel, PowerPoint, Teams, CRM etc., unless they decide to keep the older versions.
So AI tools will ultimately become omnipresent in the work life, leading to the ever pressing question about the relevance of humans, especially in banking.
Will AI take your job?
For cognitive tasks, artificial intelligence is not directly taking jobs. It cannot replace white collars, but instead will augment their intelligence and efficiency. It will modify the nature of work such that what people do at work and how they do it will radically change. With two consequences: By dramatically increasing the productivity of employees, fewer of them will inevitably be needed. And those who remain will be the ones having best taken advantage of AI powered tools.
The widening gap
And now, a side reflection on which I will come back in a later article. I am amazed by the increasing gap between the level of use of digitisation and now of AI at the personal level, and the old fashioned way how companies deal with each-other in the B2B sphere. At some stage, something will happen to the latter which could be quite disruptive.
Comments