To navigate through the worldwide Covid-19 health crisis, organisations from every sector have needed to adapt enterprise and customer-facing operations using AI, the cloud and other technologies, to stay relevant for employee and consumer behaviours. The shift to remote working has meant that plans for digital innovation needed to be accelerated, and AI has been key to easing transitions and improving customer experience.
Initially, fears were frequently expressed about the evolution of AI meaning that human workers would no longer be needed. But with flaws such as bias and inaccuracies remaining common, the contrary has proven to be the case.
“If you’ve watched any sci-fi film or TV programme, you’ll be aware of the fears around AI.,” said Samantha Humphries, senior security specialist at Exabeam. “Most commonly, this centres on AI becoming self-aware, overruling humans, and eventually rendering us obsolete. But, in the words of HAL 9000 in Space Odyssey: ‘I’m sorry Dave, I’m afraid I can’t do that.’
“These concerns have become so widespread that it’s created a sort of moral panic around the term. Needless to say, these concerns are for the most part unfounded. AI today offers huge practical benefits – both in terms of a vastly improved customer experience and as a support mechanism for the entire workforce.
“The power and potential of AI may seem scary, but by helping operational staff to do a better job, more accurately and more efficiently the technology is providing huge value across all industries. There will always be controls around AI – there has to be. Artificial intelligence is actually a misnomer; humans control AI and that will always be the case. The technology is there to support us, not control us.”
This article will explore some of the most prominent cases of AI transforming enterprise operations as workforces continue to work from home, and improving customer experience.
One area of the enterprise that’s seen notable transformation is HR. With organisations continuing to hire, onboard and retain the best possible talent, AI has helped HR departments to facilitate these operations remotely, while monitoring the experience of current staff.
Data insights can be used to train models with the goals of keeping employees engaged and productive, finding job candidates that are the best fit for the role and organisation, and onboarding those new hires with minimal disruption.
Malcolm Ross, vice-president of product strategy at Appian, explained how AI capabilities has worked its way into HR operations, and made them more efficient: “We now see many consumer-facing AI technologies permeating into back office and enterprise operations, such as intelligent chat bots for HR to employee communications.
“AI done right is transparent. Organisations seeking transformative impacts need to first get a hold of their overall workflows and processes, and then decide where AI decisions can be injected to deliver the best results.
“For example, an organisation can look at a recruiting process and identify a point in the process where initial resume and work history is evaluated. That decision can be augmented by AI services using natural language processing and entity analysis to assist HR recruiters to prioritise possible candidates.”
What trends are driving tech recruitment in the UK and US?
As the range of ultilised devices for shopping has widened during the pandemic, so has chosen channels, meaning that brands need to be accessible using any possible method, and have the right information in real time. Hosts of customer-to-brand conversations in today’s world include the company’s website, social media and over the phone.
According to Mark K. Smith, ContactEngine’s CEO, “half the time there is already a known problem, i.e. it was not ‘right first time’” when it comes to interactions, but “proactive conversational AI fixes that.”
Smith continued: “Nearly all companies hold the necessary data for ‘person-based’ comms. ContactEngine uses propriety proactive conversational AI to keep people informed and listens to their replies and understands what they need in over 90% of instances. This is a revolution that has only very recently become possible as the ‘transformer models’ used in AI (or, more specifically, in the NLU subset of AI) themselves come of age.
“Working from home is now just a thing we do and instant responses and swift answers are the new normal in speaking with brands – but in truth, brands cannot use humans to do this – there simply are not enough of them, and it is simply too expensive, so for the vast majority of ‘customer to brand’ conversations the computer is best.”
Information Age roundtable: how to successfully engage customers post-pandemic
Addressing identity theft and fraud
AI has also been used within sectors such as insurance and banking to combat possible fraud. Using customer data, machine learning models can comprehend biometrics, account details and behavioural information in order to determine legitimate queries.
“Over the last few years, but in 2020 in particular, AI has demonstrated its value in protecting enterprise customers from identity theft and improving their service experience,” said Claire Woodcock, senior product manager at Onfido.
“Just as the pandemic led many enterprises to accelerate their plans for digital transformation, there was a notable shift towards online identity attacks from fraudsters.”
Research from Onfido found that identity document fraud saw a 41% increase over the course of 2020, with many first-time fraudsters tried their luck to shield from the economic downturn. Attackers were found to have used 2D or 3D masks to bypass selfie and video verification processes, while deep-fakes rose in popularity. But Woodcock said that AI “has excelled for many enterprises” in finding fraudsters without compromising the customer experience.
“AI’s ability to learn from the changing risk landscape and respond with new measures and checks, coupled with manual intervention where needed, means that enterprises can stay one step ahead and keep the risk of fraud low,” Woodcock added.
“In turn, this means legitimate customers do not fear getting their identity stolen and can be onboarded quickly and securely.”
How combining AI and humans can help to tackle cyber fraud
In addition, AI has transformed fraud detection operations in the insurance sector, with models being trained to help make quicker decisions when account holders make claims.
“Instead of a customer calling up a call handler to say they’ve had an accident, and the handler deciding whether to write the car off or repair it, machine learning algorithms work behind the scenes to inform the call handler.
“This is based upon information from the account holder, the value of the car, the cost for repair, and damage that’s been caused. If you can get an immediate answer, that makes the customer experience much better than the customer otherwise waiting 30 to 60 days.”
Predictive medical diagnoses
Another area of rising AI utilisation has been in healthcare, with machine learning being used to facilitate earlier predictions and aid diagnosis screenings, including finding signs of cancer.
“In the healthcare sector, where unstructured data is common, utilisation of AI has increased dramatically in the last five to 10 years,” said Brown.
“AI is being used to better understand the root cause of problems through MRI scans or CT scans, as well as patient records, to find abnormal behaviour.”
AI use cases in healthcare for Covid-19 and beyond
Kheiron Medical, for example, has an AI-powered breast screening solution called Mia (Mammography Intelligent Assessment), which radiologists can leverage in order to detect signs of cancer more quickly.
Sarah Kerruish, chief strategy officer at Kheiron, explained: “Breast cancer detection is one of the hardest detection tasks, and calls for two radiologists. If those two radiologists disagree, then it goes to a third reader who arbitrates that decision.
“The fundamental decision that radiologists make is either ‘recall’ or ‘no recall’, so we designed Mia to work in that screening context to make that same decision. With the sector dealing with a shortage of breast radiologists, Mia can come in to be a second reader if needed, but that decision still goes to arbitration if the two parties disagree, so there’s still that human in the loop.
“By doing very careful, rigorous trials, with large-scale data sets, we’ve been able to prove that machine learning can identify mammograms where there is a high likelihood that there is cancer present.”