Sunday, June 16, 2024

Exclusive: ‘A is for augmented’ – Bjorn Austraat, Wells Fargo in “The Fintech Magazine”

Artificial intelligence is becoming ‘part of the business fabric’ at Wells Fargo, according to its Senior Vice President, AI Enterprise Solutions, Bjorn Austraat – a utility as essential and anonymous as electricity. For now, at least 

For decades, science fiction has warned that artificial intelligence (AI) is on a crash-course with humanity: that minds superior to our own will inevitably wage war against their creators, coding inhumanity into the superhuman brains of the future.

But there’s always been a parallel discourse – subtler, perhaps, than the space operas and rampaging robots of the silver screen – which imagines humans and machines forging a closer and closer union. These futurists argue that, following this theory, we’d combine ever-tighter into ‘cybernetic organisms’. The human would evolve, inexorably, into the cyborg.

Even as certain experts grumble that AI has dramatically under-delivered on its hype, it’s difficult to deny that we’re on our way to cyborg sentience in our communication, our consumption and our digital-first workplaces. But we’re not there quite yet, according to Bjorn Austraat, whose position as senior vice president of AI enterprise solutions at Wells Fargo gives him a panoramic view of AI’s current and future applications.

Like a diplomat on the Star Trek Enterprise, Austraat’s role sees him mediate between organic brains and cybernetic minds finding ways in which they can collude and collide productively.

“Many years ago, in my very first job in Austria, I was an interpreter – working at conferences, at the United Nations, as a translator,” he says. “Back then, I helped people understand each other; today, I jokingly say I have the same job, just that now it’s focussed on translating between the languages of business, data science and engineering.  And those languages and cultures are quite different. The challenge comes in tying them together: getting them aligned, balanced and moving in the same direction.”

Contrary to fears that AI is set to eclipse the need for human intelligence in the workplace, Austraat sees its smartest application – in banks and elsewhere – as working in synergy with our own minds: an alliance of computer IQ and mankind’s EQ.

“AI is really best used when it’s augmenting, so we speak of augmented intelligence, rather than artificial intelligence,” he says. “Because it’s augmenting people, it frees them up from very routine work or things that are simply not possible.

“AI is very poor at having empathy or common sense, or creativity. So those are the qualities that we absolutely want to free up humans, from their routine work, to do – the creative and uniquely human-suited work. For instance, if somebody called into a bank and said ‘I’m going to get married’, the correct response would be ‘wonderful, I’m so happy for you’ but that’s something that is not accessible to machines. We could fake it, but it’s really not the domain of AI to be empathetic.”

On such a call, AI is best left to hum innocently and productively in the background, leaving a human operative to think creatively about rapport building responses. That’s exactly what one recent AI solution enables at Wells Fargo – a cyborg-spirited innovation that the firm calls Advanced Listening. Monitoring all of Wells Fargo’s customer communication channels, Advanced Listening AI can guide customer service representatives to respond intelligently to customer queries, and can understand and categorise the gist of every interaction.

Then algorithmic models look for changes and emerging trends to provide a better customer experience. The bank’s Complaints Data, Analytics, and Reporting (CDAR) team, which sits within the Enterprise Complaints Management Office, established in 2019, has developed Advanced Listening tools and processes that leverage natural language and speech recognition. CDAR also helps to identify new complaint trends and emerging risks, along with performing diagnostic analysis to identify potential systemic issues.

That is critical for a bank that’s worked hard to regain the trust of both customers and the regulator in the wake of its invalid accounts scandal. For a period of 14 years from 2002, the firm’s targets-based sales approach saw Wells Fargo staff issue millions of unverified accounts to customers without their knowledge or consent.

The scandal, which has cost it close to $3billion in court actions, spoke to the weak spots in the human mind, and some inevitable pitfalls of human management. It’s difficult to imagine a team-up of man and machine making the same mistakes. Avoiding those sorts of consequential costs would probably swell the $447billion that Autonomous Next research recently predicted as being the aggregate potential savings from AI deployed in the banking sector by 2023.

AI promises to co-create a new era of productivity, compliance and customer satisfaction at Wells Fargo by sitting alongside Austraat’s colleagues.

“AI is really becoming part of the business fabric at Wells Fargo the era of early experimentation is over and now it’s being infused into a lot of processes sometimes as a utility, sometimes as a driver and a leader of transformation,” says Austraat.

“Our AI projects span everything from machine learning to natural language processing, customer experience, team member experience, fraud prevention, the deposits team, and much more.”

Pioneering approach to AI

Wells Fargo famously sports the logo of a six-horse stagecoach, trailblazing into the American west, a throwback to the country’s Gold Rush, more than 160 years ago. That trailblazing is now focussed on making use of financial data to build AI models that work hand in hand with human clerks. But Austraat and his team are keen to avoid the hammer-nail trap of green-lighting every AI project that sweeps across their desks.

“When people come to our team with an AI idea, the very first question we ask is ‘is that a good idea for AI?’. And sometimes the answer is ‘yes, absolutely, this is the sweet spot for AI, it can do really well there’, but sometimes the answer is ‘no, this should either be solved with robotic process automation (RPA) or an Excel spreadsheet, or rules’. We can write interesting code an interesting science experiment but, at the end of the day, it has to deliver value for the enterprise.”

That was the logic behind Wells Fargo’s creation, in-house, of its predictive analytics tool, servicing mobile banking customers with AI-generated spending reports and savings recommendations.

The bank has also previously worked with chatbot specialist Kasisto, which joined the six-month Wells Fargo Startup Accelerator programme in 2014. While in the programme, business leaders within Wells Fargo provided mentorship and guidance to help refine its solutions.

Austraat is clear that AI firms offering software-as-a-service (SaaS) aren’t the whole answer for bank, though.

“Partnering is great, but it’s also really important for a highly-regulated bank to have a core competency within AI,” he says. “You need a blend – you want to tap into the innovative spirit of the very large third-party ecosystem, but, at the same time, abdication is not OK. You need to be able to stand up to the regulator and say, with good conscience, that you did everything you could to make sure that your AI is fair, unbiased, and robust.”

Austraat is also clear-sighted about where and when AI is an appropriate investment. “AI, at the end of the day is just software, math, it’s not magical,” he explains. “So, one of the roles we play within Wells Fargo involves both evangelising and having pragmatic conversations around AI. Some things it’s good at, some things it’s not very good at. We don’t have general purpose AI, à la Terminator or Commander Data on the Star Trek Enterprise. We have speciality AI that’s very, very good at one task – like reading 800,000 pages of text in seconds.”

AI is clearly nowhere near working autonomously, without human oversight. All it can do – for now, at least – is whirr in the background, awaiting tasks that human bank staff would rather outsource to task-specific algorithmic assistants.

“Ideally, AI is invisible,” Austraat says. “It’s like electricity: it just works. You don’t want to see sparks, you just want to see the work output: the toast was toasted, the light comes on. As we move into the industrialisation phase of AI, the technology should simply become a background utility.”

Far from the gun-toting Terminator, then. Instead, what Austraat is describing sounds a lot like the cyborg future first theorised in the 1960s: humans and machines so synonymous, that the distinction between them vanishes – and the technology itself becomes as invisible as the very qualities that make us human.


This article was published in The Fintech Magazine: Issue #18, Page 92-93

People In This Post

Companies In This Post

  1. Doha Bank and Mastercard Announce Long-Term Strategic Partnership to Shape Qatar’s Payments Landscape Read more
  2. eBay Launches Venmo as a Payment Option Read more
  3. CRIF Completes Bond Issue With Another US Institutional Investor Read more
  4. Payzli Announces Strategic Leadership Restructuring to Accelerate Growth Read more
  5. Bank for International Settlements and Bank of Canada launch BIS Toronto Innovation Centre Read more