By Kimberly Nevala
August 13 — The forecast regarding AI’s impact on jobs is foggy, at best. In Cognizant’s 2015 paper, The Robot and I, one out of five companies surveyed reported up to a 25 percent reduction in employees in core administrative functions including supply chain, HR and finance.
The industry vertical predicted to have the highest level of staff reduction long-term? You guessed it: banking.
More recently, while acknowledging the likelihood of short-term job loss, Gartner predicted that by 2022 AI will create more jobs (2.3 million) than it replaces (1.8 million) annually. Other opinions (many of which Georgious Petropoulos skillfully documented on Brueges.com) abound.
Yet, as divergent as many of these opinions appear at a cursory glance, they are not necessarily in conflict. Why? The question is not: will AI will replace, augment or create jobs? (It will.) The question is: when and in what proportion will each circumstance occur?
To demonstrate, let’s consider the application of AI in customer service.
Initial AI solutions focus largely on automation: deploying conversational chatbots to answer rote, high-volume customer queries (i.e., what’s my balance, transfer money between accounts and so on.) Implementation of such systems requires new skills including AI engineers and conversational designers. The education and experience required is fundamentally different than that of the customer service reps the AI solution is designed to mimic. Nor are the new roles as numerous. It is, therefore, disingenuous to suggest that deploying chatbots to autonomously respond to 2 billion inquiries will not result in a significant reduction in staff. Or that the $8B in cost and time efficiencies predicted when these applications come of age will be immediately offset by redeployment of displaced resources to new, as yet undefined, roles.
But why not? The argument could be made that agents displaced by chatbots could be retrained to assist with higher-touch calls previously handled by more experienced, senior agents. Thereby eliminating the job losses noted above. Which would be true if AI did not also have a role in enabling, as well as automating, customer service. For example, AI may pre-authenticate a customer based on their voiceprint and then route that customer to an appropriate agent based on their profile and non-traditional factors such their emotional state: how harried does the customer sound? AI may also assist the agent by displaying integrated customer highlights, providing cue cards for next best actions or drafting responses using the agent’s individual speech and writing patterns.
Human agents enabled with AI in this way become more efficient and effective: reported improvement in issue resolution times and customer satisfaction range from 10 percent to 35 percent. Creating a seamless collaboration between the experienced agent and AI solution requires changes to existing operating practices and training. However, the resultant productivity gains eliminate the need to retain previously displaced agents to meet demand. Even as customer inquiries requiring the human touch become more complex or increase in volume.
Last but not least? Creation of new roles for both AI (autonomous financial advisors) as well as humans (the AI-enabled banking concierge?) At this juncture, we know AI will play a central role in enabling human endeavor. Exactly what that future symbiosis between man and machine will entail is yet to be determined.
Regardless, as this simple example shows, AI (not unlike other emerging technologies) typically manifests in business in three primary stages.
- Automation: Using AI to automate rote, high-volume decisions and activities.
- Job impact? Replacement of low- to medium-skilled jobs by the AI solution. Addition of new highly-skilled roles to create and maintain the AI solution.
- Net effect? Negative.
- Augmentation: Augmenting and supercharging existing practices with AI.
- Job impact? Retention and retraining of some existing resources to exploit new AI-assisted processes. Continued investment in AI conversant resources from algorithm development to business management.
- Net effect? Neutral or Positive.
- Creation: Using AI to deploy new engagement models, products and service models.
- Job impact? Creation of new roles, the specifics of which are TBD.
- Net effect? Positive.
So how will AI impact your business specifically? It depends on the AI application at hand. Feeling overwhelmed by the possibilities? Don’t be. Considering the long-term implications of AI on business and society is an important but complex undertaking. Assessing the material impact of AI on your workforce within the context of your business strategy need not be. Simple questions to ascertain whether existing jobs are likely to be axed, augmented or added as a result of proposed AI solutions include:
- Which tasks are repetitive, high-volume and well-defined?
- What percentage of an individual’s role do these tasks comprise?
- Which activities and decision points can be streamlined? Eliminated?
- What information, provided when, could expedite an individual’s decision or action?
- How much individual bandwidth will be created as a result?
- What is required to create, monitor and maintain the AI solution?
Last but not least: while AI is a hot topic by any standard, multiple studies (such as this recent survey of North America’s biggest banks) confirm that most companies are just beginning their AI journey. As a result, the most immediate impact on jobs may well result from seemingly mundane predictive analytics applications, no AI required! As well as rapidly evolving FinTech solutions as digital innovators reinvent how individuals manage their money, make payments and invest with and without AI.
As the Director of Business Strategies for SAS Best Practices Kimberly balances forward-thinking with real-world perspectives on business analytics, data governance, analytic cultures and change management. Kimberly’s current focus is helping customers understand both the business potential and practical implications of artificial intelligence (AI) and machine learning (ML).