It’s a year since the launch of ChatGPT, an event that marked an inflection point in AI adoption. This innovation catapulted AI from a futuristic concept to an everyday reality. As the speed and sophistication of AI tools increase, we must ask ourselves: Where does human agency sit in an AI-enabled future?
To answer this question, consider a lesson from Betty Crocker. When the iconic instant cake mix was launched in the 1950s, suddenly home bakers needed only to add water and stir. The process was virtually effortless and foolproof, producing the perfect sponge cake in minutes. Nevertheless, many US housewives were not keen.
In a bid to boost sales, the brand brought in consumer psychologist Ernest Dichter, who uncovered the problem: the cake mix was so simple to use that housewives felt guilty for doing so. So, Betty Crocker changed the recipe to ask bakers to add fresh eggs, an extra step that, while elementary, added a “personal touch” and changed how customers perceived and valued the brand.
The story demonstrates the reluctance to dispense with human effort and a counterintuitive desire to retain some personal involvement in process (even if less convenient than letting the AI do its work unaided), a phenomenon now known as the “Betty Crocker Effect.” Today’s AI developers should take note: full automation is not always the first choice for customers, even if it’s the easiest, cheapest or most efficient option.
For businesses, the attractions of AI are obvious. Almost 85% of top executives believe it will give their organizations a competitive edge, whether that’s through reduced costs, increased efficiency, improved customer service, better forecasting or faster decision-making. As this technology continues to gain traction, the global AI market is expected to reach up to $2tn by 2030.
But in our haste to adopt AI, we must not forget the value of the human touch. While some advanced AI systems may replicate the human decision-making process, they may never access the subjective insights and instincts accumulated through years of lived experience.
Imagine a machine learning (ML) algorithm programmed to manage the supply chain of an industrial manufacturer. The program could recommend a particular driver, route and vehicle to transport materials from factory to warehouse, considering real-time factors such as traffic, logistical capacity and current weather conditions. In most cases, it would be able to identify the most efficient option, relying on data metrics alone.
However, the AI model has inbuilt limitations. For instance, it can’t account for unforeseen variables that deviate from historical data, or the socio-cultural factors that influence situations – a driver who’s likely fasting for Ramadan, for example, or a flash flood causing congestion in a nearby town. The model can handle known patterns and scenarios, but it lacks the data to adapt to sudden changes. Nor can it modulate the decision-making process to consider more intuitive influences, such as a driver returning from a long lay-off owing to illness.
This is where human decision-makers add crucial context, incorporating knowledge of local conditions, customer preferences, and the typical (or atypical) behavior of individual drivers. Those of us blessed with the intuition and capacity for innovative thinking that machines lack can respond creatively to dynamic environments. While, at first glance, this may not seem like much of an advantage over the huge processing capacity of AI, when small decisions become the catalyst for larger ones, it can make a significant difference.
Even if an AI model has all the quantitative data required to make a decision, human intervention may still be necessary to consider the moral or ethical implications of that decision. For example, an algorithm can’t account for the impact of automation on the morale and performance of the human workforce it is partially displacing, as well as the remaining human workers. Similarly, it will struggle with emotional concepts such as fairness, empathy and societal bias, which could impact decisions related to hiring or financial lending.
The prevalence of bias in some AI recruitment tools is a concerning issue that underscores the need for human oversight of these processes. An algorithm is only as diverse as the data that scientists use to feed it. If biased legacy hiring practices are fed into an algorithm, it can unintentionally carry over this discrimination from the past, and in some cases even amplify it.
It’s clear that many situations call for an integrative approach to AI that supplements technology with a human in the loop, retaining the oversight and guidance that only people can provide. But, in practice, what would that look like?
While AI offers numerous advantages, it is essential to avoid over-reliance. Businesses should be selective about the tasks they assign to AI systems, ideally reserving it for repetitive, rule-based tasks where the cost of failure is low, and the benefits and scalability of automation high.
Companies should review the contribution of their human workforce. What insights do they bring to the table? How can leaders incentivize this value-added uniquely “human” behavior? By considering these questions while patterns are still incipient, organizations will be able to ensure their staff continue to feel valued in the human-AI partnerships that develop.
For some firms, the changes will be significant. Certain operational processes may require restructuring to allow people to work in harmony with AI. Other workflows may be rewritten entirely to include AI by design. Be prepared to adapt your strategy continuously based on feedback, changing business needs, and further advances in AI technology.
Transparency is another key ingredient of achieving an AI transformation free from hidden error and bias. Staff should understand exactly how the algorithm works so they can explain the decisions it makes, and identify situations where human judgment is required.
Our world may be built on data, but some outcomes can only be reached through the application of emotional intelligence – a quality that even the most sophisticated AI has yet to develop.
This will require companies to retrain their employees, both to understand AI and to enhance their skills in areas that complement AI’s capabilities. As with any new technology, many integration hurdles can be mitigated through education and appropriate communication, and companies should take care to position AI as a tool to enhance human potential, rather than replace it. Making the effort to demystify and democratize AI across the workforce will go a long way towards addressing the mistrust that many employees feel towards this technology.
Additionally, companies that are preparing to introduce customer- or client-facing AI models will need to consider how they communicate this decision externally. Many customers are as wary of such changes as doubters in the workforce, and might feel that a reduction in human involvement will see a reduction in quality of service. Management should take care to highlight incentives for the individual, including shorter wait times and lower costs, while emphasizing remaining human touchpoints.
Steps like these will help smooth the transition to an AI-enabled future. Companies will see the best results from AI when they complement its use with soft skills that are uniquely human. Workers will find it easier to embrace this technology if they can understand how they will be integrated in its roll-out.
Exciting new opportunities are emerging in the field of AI as some advanced models surpass humans in labor-intensive tasks that require immediate access to a huge dataset, such as pattern recognition, fraud detection and predictive analysis. However, there is always the caveat that this technology is not without risks or limitations.
It is essential that, as the capabilities of AI grow, we continue to remember its shortcomings. It is unlikely that AI will ever be sufficiently sophisticated to capture the wealth of actively and passively acquired knowledge that informs human decision-making. It is essential to combine ML with human expertise, taking a holistic approach that considers both quantitative and qualitative factors.
The Betty Crocker Effect shows there will always be a place for human contributions; few people are ready to lose that personal touch altogether. Our world may be built on data, but some outcomes can only be reached through the application of emotional intelligence – a quality that even the most sophisticated AI has yet to develop.
A flurry of venture capital activity presents investors with a challenge: how to distinguish genuine innovators from opportunistic players capitalizing on the AI hype. Enter Harshul Sanghi, a seasoned technology investor with...
26 April 2024 • by José Parra Moyano in Artificial Intelligence
AI holds incredible promise to improve virtually every aspect of our lives, but we can't ignore its risks, mishaps, and misuses, author Art Kleiner tells José Parra Moyano....
23 April 2024 • by Jerry Davis in Artificial Intelligence
GenAI can deliver cost and productivity benefits for colleges and students, but there is a dark side....
Audio available
22 April 2024 • by José Parra Moyano, Christine Legner, Konrad Schulte in Artificial Intelligence
Training and fine-tuning AI models requires troves of data, but storing and processing that data is financially and environmentally costly. Here are three ways to address the problem....
Audio availableExplore first person business intelligence from top minds curated for a global executive audience