In a white paper released in October called Digital Transformation in Financial Services, I described Digital Transformation as the profound and accelerating transformation of the customer experience, business model and operating model brought about by the application of digital technologies. In a subsequent whitepaper written in November – Artificial Intelligence: Current Uses and Limitation – I explained how organisations are deploying A.I. throughout their business processes. Both papers discuss how emerging technologies are driving innovation across industries but what they also have in common is to highlight the fact that the involvement of people is key to any successful implementation
Instead of merely replacing human labour with machines and automated processes, Digital Transformation and A.I. are most valuable when their deployment enables individuals, teams, and entire organisations to perform more efficiently, to enhance existing processes, to adapt quickly to changing business environments, and to generate new business channels. Digital Transformation may be driven by multiple emerging digital technologies but instead of using them in isolation, it is about enabling innovation, customer focus and rapid innovation, and developing the appropriate human skill sets to support these goals. Similarly, A.I., in its current state, is best suited to provide insight into large data sets due to its ability to find patterns in them through highly complex calculations or to automate predictable, repetitive tasks. Human abilities such as critical reasoning, emotional intelligence, creativity, inspiration, and wisdom cannot yet be well replicated by machines but when combined with A.I. they form a collective intelligence that highlights the power of not simply replacing humans with machines, but in supplementing humans with machines.
The Human Role in A.I.
In the healthcare industry, great strides have been made through the use of A.I. in the early detection and/or diagnosis of illnesses as well as recommending successful courses of treatment. Rather than replace the doctor, however, A.I. is used to augment their knowledge through the analysis of historical data in order to find patterns that help them to reach the best possible decision in conjunction with their own experience. On occasion, A.I. will provide an output that the doctor disagrees with or is plain wrong. In this case, human intervention by the doctor can guard against mistakes.
Aside from partnering with A.I. to form a higher level of collective intelligence, humans must also play a role in the use of A.I. by evaluating opportunities for its use, by supervising and managing it, and by implementing it. This, in turn, means that there is a human responsibility in ensuring its ethical and appropriate use. Firms should care about it too in order to help overcome bias or negative reactions to A.I. and to help accelerate its use.
Machine Learning algorithms used in A.I. are developed and fine-tuned by learning from experience – otherwise known as data – and being trained (either by itself or manually) to reach ever more statistically accurate decisions. The fact that the decisions they make are based on the data fed into them means that the outcome of those decisions can be influenced when selecting the data used. If a Machine Learning algorithm is trained on biased data it will generate biased outcomes.
Network architectures of algorithms can also be set up so that some data points are weighted in favour of others. This too can generate biased outcomes. This may be done deliberately based on qualitative assumptions (e.g. a convicted murderer poses a greater threat to society than someone charged with littering and should, therefore, be flagged more heavily on a police database), but as soon as human opinion starts to play a part in the decision-making process, even for presumably well-intentioned reasons, the impartiality of the outputs generated can be compromised.
Without human interaction acting as a guardrail and left to its own devices, there is the potential for A.I. to make the wrong decisions or to make decisions that can’t be explained due to a lack of traceability in how it reached such a decision. This is one reason why people weary of the potential negative impact and safety of A.I. worry about its use. So-called black box solutions – A.I. tools developed by providers to train on either user or proprietary data without disclosing the underlying network architecture – could be open to untraceable and unexplained manipulation and abuse. The focus of much of the current research around A.I. is around interpretability so that machines can explain decisions they make, but this is not yet available and as such, is an important consideration for people to make when deciding how to use A.I..
Data is the fuel on which A.I. runs and, in general, the more data it consumes the greater insights it can provide. As users of A.I. seek to consume greater volumes of data, the implications around where the data comes from and who owns it grow. There are substantial benefits to be gained for firms that can use A.I. to fine-tune their offerings and customer interactions to individual tastes, including boosting sales and revenues through the enhancement of the customer experience. A plethora of data is now available through the everyday digital interactions that people have through their smartphones, computers, and internet of things including personal data collected through apps, social media, and private photos. Much of this may not have been intended for outside use, however, and firms need to be transparent about where they source the data they use and how they use it.
The solution is to form policy that ensures its appropriate and ethical use and potential revisions to government legislation around data privacy laws and access rights. Some of the larger providers, such as Google’s DeepMind have already started the conversation around this and although still in its infancy, it is another example of the new sorts of roles and responsibilities being created by A.I. that will need to be policed by humans.
A.I. Impact on Jobs
There is no hiding from the fact that some existing jobs will be automated as technology advances, however, this is not a new phenomenon and is not unique to A.I.. A.I. can boost productivity; it doesn’t get tired, it lowers risk by removing human error, and it lowers costs as some human jobs are replaced by machines. Advocates of A.I., however, argue that by using it to automate certain low-level tasks it can free up human workers to focus on more interesting, value-add activities. Indeed, a study conducted by the McKinsey Global Institute demonstrates that whilst most industries have tasks such as data collection and processing that can be automated, they also have other tasks like managing and interfacing that cannot be. This means that in many cases, only specific tasks rather than entire jobs will be handed over to machines.
In addition to the requirements for human ownership in implementing and policing it, A.I. could also create new jobs through both short-term bubbles and the longer-term development of new business channels. Much Machine Learning still requires manual and onerous tagging and training to make it work and this has already resulted in the creation of employment and earning opportunities through services such as Mechanical Turk, Upwork, and Freelancer. Ten years ago, nobody knew they wanted an iPhone, yet today the product line makes up around 65% of Apple’s revenues, a firm that now employs nearly 120,000 people worldwide versus just over 20,000 when it was first released in 2007. This is evidence that in the long-term, products and industries that we haven’t even thought of will produce new jobs.
The Human Role in Digital Transformation
Digital Transformation is a phrase that is often misunderstood. It has been driven by digital technologies such as social media, online applications, mobile devices, cloud computing and A.I. that generate, store and process data, but it is fundamentally about implementing new processes to create, enable, manage
and deliver digital products and services with a continuous and iterative approach to integrating business, operations, and emerging technologies. That requires organisational change and strong top-down leadership to foster and drive a change in mindsets and therefore involves significant human input.
Successful Digital Transformation can see revenues grow through the creation of new products and increased productivity
Firms that have prospered in their programmes of Digital Transformation have fostered a culture of organisation-wide collaboration between employees and empowered them to be creative whilst failing fast. Rather than focusing on merely cutting cost, successful Digital Transformation can see revenues grow through the creation of new products and increased productivity. New skills are required to implement modern processes and transformation programmes that redefine how firms operate. Firms that are defensive in their business strategy by focusing on cost leadership may actually be better served by directing resources towards building new business models and developing new product lines.
Successful Digital Transformation requires a flexible workforce that can switch seamlessly and intuitively between tasks and the combination of the right technologies and well trained human labour is fundamental to that.
Conclusion
Digital Transformation and Artificial Intelligence are two driving forces behind modern business innovation and to be effective they each require the right mix of human and technology input. Today’s A.I. is regarded as “narrow A.I.” and is built with a singular purpose without the ability to fundamentally change the scope or boundaries of its design. Artificial General Intelligence, where machine intelligence has full human-like capabilities and can perform any task that a human can do, is still many years away. Until machines are capable of doing everything that humans do, we will remain as the innovators, policymakers, decision-makers, managers and more. In 30 years, jobs and firms will look very different from today, and whilst upskilling, retraining and education of workers will be necessary in order to adapt to the change, it is no different to what happened during the first and the second industrial revolutions and is nothing we haven’t seen before. As John F. Kennedy stated in 1962, “If men have the talent to invent new machines that put men out of work, they have the talent to put those men back to work”.