Part III - Implications of GenAI

In the first two instalments of this series on GenAI for the layperson, I developed some foundational building blocks in relation to key terminology and relationships in GenAI (Part I - Explanation of GenAI) and then explored the emerging applications (Part II – Applications of GenAI). In this final instalment in this GenAI series, I shift my attention to exploring some of the implications of GenAi from the perspective of various stakeholder groups.

Widely diverging views on implications of the technology and its ultimate end state

Given the dramatic emergence of GenAI into the consciousness of the average person and the rapid evolution of its capabilities, there remains a broad range of views on the end states for the technology. On the optimistic end of the spectrum, the ability to automate and augment increasingly services biased modern economies, brings the potential for significant uplifts in productivity; releasing labour and capital to pursue more beneficial uses. On the more pessimistic end of the spectrum, humans are unleashing a technology that can learn and transfer information quicker (instantaneously) and more effectively than humans will ever be capable of bringing the associated risk of loss of control.

Technology has been transforming the nature of work for different groups of workers

However, in many respects, technology has been changing the anatomy of work for decades. Over the years, machines have given human workers various “superpowers”; for instance, industrial-age machines enabled workers to accomplish physical tasks beyond the capabilities of their own bodies – thereby transforming the nature of work for lower skilled labour. More recently, computers have enabled knowledge workers to perform calculations that would have taken years to do manually. GenAI is extending this impact and transforming the nature of work for higher wage white collar knowledge workers mostly living in metropolitan areas.

These examples illustrate how technology can augment work through the automation of individual activities that workers would have otherwise had to do themselves. At a conceptual level, the application of GenAI may follow the same pattern in the modern workplace, although as we explored in Part ii of this series, the types of activities occupations that GenAI could affect will vary.

The adoption process for the GenAI will be challenging and traditional AI still holds majority of value

It is worth remembering that the era of GenAI is just beginning. Excitement over this technology is palpable, and early pilots are compelling. But a full realisation of the technology’s benefits will take time, and leaders in business and society still have considerable challenges to address. These include managing the risks inherent in GenAI, determining what new skills and capabilities the workforce will need, and rethinking core business processes such as retraining and developing new skills.

Is likely that the other applications of AI (not GenAI) will continue to account for most of the overall potential value of AI. Traditional advanced-analytics and machine learning algorithms are highly effective at performing numerical and optimisation tasks such as predictive modelling, and they continue to find new applications in a wide range of industries. However, as GenAI continues to develop and mature, it has the potential to open wholly new frontiers in creativity and innovation.

Democratised open access and equal benefits to good and bad actors are important characteristics

There are two inherent characteristics of the technology that play an important role in thinking about GenAi’s implications to various stakeholder groups. Firstly, is the technology’s open and relatively low-cost access. Unlike other powerful technologies, such as nuclear power, where access is generally closed and high cost, GenAI capability is broadly accessible at low to zero incremental cost. Secondly, the same inherent technological capabilities that underpins so many positive use cases also equips bad actors in equal measure. We are already getting glimpses of this negative potential in deep fake applications and the next generation of scam emails starting to emerge.

Differential distribution of power in society can distort the relative winners and losers

Despite GenAI’s open access, history suggests that new technologies don’t necessarily bestow the same benefits to all stakeholders. Relative distribution of societal power plays an important role in determining the dispersion of benefits and costs across society. For example, while the introduction of cotton gin processing technology bought much benefit to landowners in the early 19th century USA, it did little to uplift the life experiences of the black slave labour working the fields. Therefore, GenAI unleashes the possibility that the gulf between the haves and have nots increases.

GenAI’s negative potential impact on trust and truth can have material consequences

Equally, GenAI has the potential to seriously dent trust; one of the baseline fabrics of community and societal life. While this can come in overt ways through the few actions of bad actors it is just as likely to come from the iterative actions from many creators (all the way up to the President of the USA!) of a virtual world that becomes increasingly indistinguishable from the real world. Few technologies in the past have had the ability to blur detection of the truth to the extent that GenAI can. In addition, when you appreciate that GenAI has absorbed every book ever written on persuasion and is increasingly showing an ability to create intimate relationships with users, you don’t need to believe in out-of-control robots to wonder what sort of society may evolve over time.

GenAI demands a multi-functional organisational approach

While many organisations approached traditional AI through siloed experiments, GenAI requires a more deliberate and coordinated approach given its unique risk considerations and the ability of foundation models to underpin multiple use cases across an organisation. For example, a model fine-tuned using proprietary material to reflect the enterprise’s brand identity could be deployed across several use cases (for example, generating personalised marketing campaigns and product descriptions) and business functions, such as product development and marketing. To that end, it is even more important to convene a cross-functional group of the company’s leaders (for example, representing data science, engineering, legal, cybersecurity, marketing, design, and other business functions). Such a group can not only help identify and prioritise the highest-value use cases but also enable coordinated and safe implementation across the organisation.

Data infrastructure immaturity may be a handbrake on organisation adoption rates

Companies will also have to assess whether they have the necessary technical expertise, technology and data architecture, operating model, and risk management processes that some of the more transformative implementations of GenAI will require. For example, the lifeblood of GenAI is fluid access to data honed for a specific business context or problem. Companies that have not yet found ways to effectively harmonise and provide ready access to their data will be unable to fine-tune GenAI to unlock more of its potentially transformative uses. Equally important is to design a scalable data architecture?that includes data governance?and security procedures.

Responsible AI has to be addressed at macro and micro levels

This raises the issue of responsible GenAI. GenAI brings renewed attention to many of the same risks associated with broader AI, such as the potential to perpetuate bias hidden in training data and managing data security. It also presents new ones, such as its propensity to “hallucinate”, infringe on intellectual property (IP) rights, and produce different answers to the same prompts thereby challenging reliability. Large language models can be prone to “hallucination,” or answering questions with plausible but untrue assertions. Additionally, the underlying reasoning or sources for a response are not always provided. This means organisations should be careful of integrating GenAI without human oversight in applications where errors can cause harm or where explainability?is needed. It will also be interesting to see how society reconciles the energy intensity of GenAI, and other digital technologies such as crypto for that matter, with the vigorous pursuit to zero carbon.

Adoption rates and released societal benefits remain unclear with important pre-requisites

The demands of responsible GenAI in an environment of declining societal trust and in the midst of the inevitable tug of war between good and bad actors are likely to result in adaption rates at a collective level taking longer than the initial hype would otherwise suggest.

Like many of the technologies before it, the most common value case of GenAI is likely to liberate time by enabling processes to be completed quicker and better. However, for living standards to improve off the back of improved productivity, capturing the released time for more productive uses is a pre-requisite. This is not a predetermined outcome. For example, in the last decade productivity in Australia has grown at its slowest rate in 70 years despite the bulk of the population having more technology in their hands (iphone) than prior generations had to travel to, and land on, the moon.

So as I draw this three-part series on GenAI to a close, I finish up reflecting on the very strong and divergent views that key commentators have on the impact of the technology on our humanity. As science did to religion, technology is increasingly impinging on the territory of traditional human activity (work and interactions). How far this will go, how effective this will be, and whether it unlocks new capability in human activity remains uncertain. In the defense of humanity and its virtues many commentators take the best characteristics of humanity for the technology comparison. However, like machines, the same foundation blocks for humanity can be (and are being) used for immense good and for immense bad. In reality, humanity is far from a perfect comparator. It is for this reason that there is no universally ‘right’ answer and the tug of war between humanity and machines is likely to be a lively one for many years to come.

Thanks for joining me on this journey of enquiry and insight into GenAi. I trust that it has equipped you with better understanding of key terms and definitions, stimulated thinking of where application use cases are more likely to occur and challenge a range of implications of the technology for society, business and as individuals. I welcome your engagement to share and build on some of the topic areas covered in the spirit of building understanding and refining perspectives.

Andrew Rosengren MD Guberno Consulting