COVID-19 upended a lot of things, and among those were artificial intelligence (AI) algorithms which were supposed to forecast the future. As it was a once in a century Black Swan event, the algos did not have past data to crunch to come up with predictions. It was then that humans and AI worked together to create an optimised model. The Boston Consulting Group (BCG) and MIT Sloan in a research found that humans and AI interacted to create a far better model than AI working by itself.
At Walmart, employees not only have extensive operational experience managing in-store product assortment but also an ability to understand context that even extensive historical data lacks — such as sudden, radical shifts due to the COVID-19 pandemic. Machine learning depends on historical data being relevant to current and future states. But when faced with COVID, the world completely changed, and AI could no longer predict the future from the past.
For assortment management, Walmart designs AI solutions to present recommendations to managers. Managers can agree, disagree, and comment to improve both the current recommendation and future recommendations. In a COVID-19 world, management disagreement with AI solutions was a critical source of new machine learning. Walmart ensures that decisions reflect the entirety of knowledge available – in databases as well as in people’s heads. Well-designed roles support mutual learning.
Different modes of Human-AI working
At toolmaker Stanley Black & Decker, image-processing algorithms monitor the quality of its tape-measure manufacturing. Cameras capture images as tape measures pass through various points of manufacture and, flag defects in real time before the company wastes additional resources on defective tapes. These AI systems work independently in real time because waiting for human input would slow the process. But humans still have a role, because as defects, at times, still warrant some actions to give additional validation, as there are sometimes grey areas. The process still involves human effort in exceptional cases, but it doesn’t have to slow the main process flow.
There isn’t just one way to structure and refine human-AI interactions. Instead, multiple modes of human-AI interaction need to be deployed. Organisations that successfully use different modes are six times more likely to attain significant financial benefits than those able to use just one or two according to the research. Broader competencies allow organisations to fit a wider variety of interaction modes to a wider variety of situations.
Take the instance the Al-decides-&-implements mode in which AI has nearly all the context and can quickly make decisions. Human involvement would only slow down an otherwise fast process. Repsol, a Spanish a fossil fuel company based in Madrid, Spain, embeds AI in its customer relationship management system to deliver real-time personalised offers, like discounts and free car washes, to consumers at its 5,000 retail service stations, with humans providing only a light layer of oversight and supervision and maintaining compliance with local regulations. crude and the operating conditions of the refinery, to recommend blending schedules for the next 30 days. Humans then decide which blending schedule to use, depending on expected global market conditions.AI generates insights, human uses them in a decision process: In this mode, inherently creative work requires human thought, but AI insights can inform the process.
AI can capture the context well and make decisions, but humans – rather than software or robotics, for instance – implement the solutions. Repsol uses this mode for AI predictive maintenance in offshore production facilities. AI identifies parts at risk of failure and schedules a maintenance review. Post-review, human operators then schedule the replacement, taking into account part availability and scheduled maintenance.
The company also uses the Human-Generates-&-AI-Evaluates mode where humans generate many hypothetical situations but rely on AI to tediously assess many complex dependencies. For example, Repsol uses digital twins of physical assets, such as wells, to simulate the consequences of possible operational changes and validate hypotheses. Using multiple engineering and operational efficiency models, managers can simulate consequences before actually changing a physical well.
The Word-of-Machine Effect
Harvard Business Review conducted an interesting experiment which they called The Word of Machine Effect in which they tested AI recommendations vs. human choice. The word-of-machine effect stems from a widespread belief that AI systems are more competent than humans in dispensing advice when utilitarian qualities are desired and are less competent when the hedonic qualities are desired. Importantly, the word-of-machine effect is based on a lay belief that does not necessarily correspond to the reality.
Researchers found that simply offering AI assistance won’t necessarily lead to more successful transactions. In fact, there are cases when AI’s suggestions and recommendations are helpful and cases when they might be detrimental. When do consumers trust the word of a machine, and when do they resist it? The research suggests that the key factor is whether consumers are focused on the functional and practical aspects of a product (its utilitarian value) or focused on the experiential and sensory aspects of a product (its hedonic value).
Two hundred passers-by in (pre-COVID-19) Boston were asked to participate in a blind market test for haircare products. Using leaflets to explain the test, we asked each person to select one of two hair product samples, one recommended by AI and the other by a human. As predicted, when passers-by were asked to focus only on utilitarian and functional attributes such as practicality, objective performance, and chemical composition, more people chose the AI-recommended sample (67%) than the one recommended by a person. When passers-by were asked to focus only on experiential and sensory attributes such as indulgence, scent, and a spa-like vibe, more people choose the human-recommended sample (58%) than the one recommended by AI.
Don’t just use AI; learn with AI
While adoption of AI continues to increase, and many organisations now use AI technologies to generate some business value, but many more continue to struggle to find gains. Many organisations are finding it difficult to build an AI foundation that rests on the right data, technology, and talent. Or they may have built this foundation, use it to churn out AI solution after solution, and yet wonder why the financial benefits are only incremental.
Significant financial benefits are likely only when organisations define multiple, effective ways for humans and AI to work and learn together. For instance, in one experiment, Harvard framed AI as augmented intelligence that enhances and supports human recommenders rather than replacing them. The AI-human hybrid recommender fared as well as the human-only recommender even when experiential and sensory considerations were important.
Most successful organisations distinguish themselves through effort and a commitment to learn with AI. They don’t just get good at working with machines; they get good at tailoring human and machine roles dynamically as situations change. They don’t facilitate machine learning; they facilitate mutual learning. They don’t just use AI; they learn with AI.