Dawn of The Robot Leaders: Replacement or Collaboration? (1)

In a recent article by Zeger van der Wal and Yifei Yan, the authors pose the question on whether “robot leaders” could outperform human politicians. In their article, they deem it probable that – following replacement of human labour in every other line of work – (political) leadership should expect to ultimately start receiving competition from robots and artificial intelligence as well. According to the authors, commentators from Asian and Anglophone nations are already arguing in favour of such robotization of politics in light of the inherently flawed nature of human leadership. The idea sounds pretty straightforward, but is it as perfect in practice?

More intelligent than human leaders

Greed, ego, emotionality, and our flawed rational understanding are all ingredients for failing human leadership. The authors of the WEF article thus argue that “While being governed by elected or appointed robots, cyborgs or algorithms may sound like science fiction, the advantages of robogov should still be obvious. Being far less hampered by ideological extremism and tunnel vision, or by the egotism and narcissistic tendencies that seem to characterize many government leaders, robogov may be able to make rational, fair, well-rounded and evidenced-based decisions.”

Using tech, we can simulate a bigger, quicker, better decision-making brain. However, AI can still flash crash our economy, or lead us down paths where we feel that we need or even can no longer understand how our societies are run. Especially if and where we allow “Robogov” to completely replace our decision-making bureaucrats, we might thus find ourselves lacking options to course correct.

Nonetheless, it might be argued that if AI is smarter than humans that we would be better off being ruled by robots anyway. But the authors are right to add that the prospect of the “deep learning” ability that should make AI more intelligent than us, is also very problematic because of “imperfect, contested, morally ambiguous or partly hidden information.”This is sometimes more obvious than at other times. And there seems no perspective on a solution to this problem because all information – as well as the potential parameters to sort this information – is subject to human understanding of the truth. Human understanding can go a long way, but is ultimately not neutral. This consequently goes for the way in which any such “Robogov” deals with data, but also for every other aspect of its programming. For example, the authors fear a gender bias as a result of the programmer population being largely male, thus building AI in the image of their world views.

Healthier than human leaders

The authors argue that machines do not tire like humans do. Which is true, although we should not exaggerate this point. For example, the “unlimited energy” of machinesis always limited to our capacity to provide said machine with continuous reserves of -indeed: - energy, as well as “clean and safe” software- and hardware capacity. That is: both software and hardware are usually sensitive to stress from usage and from external influences. As the authors correctly point out, such external influences do include data security risks. 

Similarly, the utilization of both hardware- and software based decision making infrastructures will, depending on said infrastructure capacity, also reach a maximum at some point, thus depleting the decision-making“ stamina”. If ill-programmed or maltreated, our “Robogov”-to-be might even face a burn-out, risking damage and malfunction to both its “mental” and “physical” health. So, even though machines do not exactly tire like humans do, they are also facing risks to their health that should not be overlooked. And whilst machine components are easier replaced than human components, a systemwide failure of a complex machine is harder to overcome than a systemwide failure of a human leader, or even a team of human leaders.

More competitive than human leaders

“Robogov will also be much less susceptible to corruption and unethical behaviour. For these reasons, we already entrust AI with deciding on complicated social policy issues, from qualifications for unemployment benefits to targeted services that suit differentiated citizens’ needs.”– Although perhaps best checked with some human oversight, robotization of specific public services has the potential of being largely beneficial in terms of effective and efficient policy execution. But these decisions, although complicated, are relatively simple policy applications that require only processing of available data leading towards categorization into pre-determined decision categories. No actual decisions are made. 

As both programming and machine learning are inherently flawed as a consequence of their human input, it seems unlikely that “Robogov” could ever evolve into the immaculate decision maker that some would hope it to become. And even insofar as they could: what ought we do with our future “Robogovs” if they have different intellectual sizes, or similar brainpower but different opinions? For example: given different cultural moralities and consequently Robogov’s decision parameters or even merely its informational input, it is quite conceivable that Eurobogov will come to very different decisions than Amerbogov, Rusbogov, Koreobogov, ChinAIgov, AIpan or AIndia.

But perhaps we are indeed getting ahead of ourselves when we focus on these utopian far away futures, whilst there is a hugely AI- and robotized, or at least intensively digital component to our current leadership realities already. The authors point out that human leadership is – despite of its flaws – currently undisputed by AI or robotic leadership. But also, that by their own failing "to examine, understand and master the pros and cons of data and robotics”, human leaders fail to assess the threats and chances that AI and robotization pose to our societies and the structures of power within.

More competitive than other human leaders

As an overall pattern, it seems that our increased access to and accumulation of social information is primarily used for the weaponization of knowledge, in ways of amplifying or disrupting national and international political and economic status quo and concurrently the cohesion of our societies. We could point at the intense electoral disruption through use of big-data driven voter targeting, or – as Zeger van der Wal and Yifei Yan point out – the new and intensively used mass surveillance capacities as well as the role of social media as a conduit and catalyst for the execution of popular “justice”.

As long as domestic leaders fail to update themselves to our digital age, the fate of every individual and of society at large is perilously left to the scruples of corporate and foreign interests and hyped popular concerns. Digital competence should thus easily be understood as a dimension of today and future power. This is true for states and their representatives, for electoral campaigns, for lobbying activities, business deals and so on. Generally speaking, the more intelligence-informed party in any negotiation has the upper hand.

Replacing human leadership with artificial intelligence may sound tempting, but the practise seems quite imperfect indeed. From hacking risks to system failure, and from gender bias to not understanding how decisions are made; Robogov may not turn out to be the solution to the inherently flawed nature of human leadership. Ultimately, Robogov is an extension of the human understanding of the truth and therefore subjected to the same flaws and biases. But what to think about a less radical option: The collaboration between artificial intelligence (AI) and humans? Part 2 of this reflection on AI and leadership will report on such an option.