Steering AI Equitably and Sustainably

Life without good politics, rigorous science, and trustworthy technology in the Age of AI can quickly become solitary, deprived, nasty, brutish, and brief, to paraphrase Thomas Hobbes’s “Leviathan.” Only imagination and benevolence can improve and safeguard the futures of billions of overlooked people. That future hinges on a new e-nvironmental ethics as we negotiate an innovative alliance between the natural world and synthetic life.

Humans have evolved to be the primary selective mechanism on the planet, pushing back against the blind forces of Darwinian natural selection. It is nature that must now adapt to humanity. Deep-learning algorithms are now designed to predict when a cell is destined to die, something human scientists have struggled to achieve for decades, and which remains a key endpoint in understanding neurodegenerative diseases. Despite these AI advances, the COVID-19 pandemic hangs over humanity as a tragic reminder that nature can be merciless.

There is no limit to artificial intelligence except real intelligence. At its most basic level, AI contains politics, culture, procedural and social practices, institutions, and critical infrastructure. It is not a new form of intelligence. It is an unprecedented form of human agency. There are four features of intelligent behaviour: (1) understanding the physical world, (2) persistent memory, (3) reasoned agency, and (4) planning complex actions, particularly planning hierarchically.

The world is changing according to an infinite set of random possibilities, and the only way to prepare for them is through abstraction. Humans have built their understanding of the world using hierarchies of abstract concepts like atoms, molecules, and materials. At every layer of understanding, we disregard much of the information contained in the layers below as it becomes irrelevant. Humans make sense of the physical world by constructing hierarchies of abstractions, much like the periodic table of elements in chemistry. AI cannot construct knowledge in these wafer-thin layers of abstraction within hierarchies.

If we train an AI on the poetry of Caribbean poets or the stories of West Indian writers, it might create different hybrids of authors and genres. The same applies to sound technology, cinematography, and fine art. Since we often judge many creations inspired by or derived from previous works as creative, it is not unreasonable to see AI as creative under certain circumstances. AI assemblages both mirror and produce social relations and meanings of our life-worlds. More AI means more human life spent more intelligently.

Computational reasoning and embodied work are inextricably intertwined. The horizon is filled with the potential for interesting collaborations between creative humans and machines in the future, in literature, finance, sound technology, film, fashion, e-commerce, medical sciences, animation, gaming, and the visual and performing arts. While these prospects enhance human life, the threats posed by cartel offenses, insider dealing, collusion, price fixing, and market manipulation necessitate a new jurisprudence. Simulation-based models of markets using artificial trading agents have demonstrated that an artificial agent can learn the technique of order-book spoofing.

Social bots are effective tools for pump-and-dump schemes. These bots can circulate misinformation and disinformation about a little-traded company to create a surge in its stock value. While such social media attempts may not sway most traders, it is precisely this type of network effect that algorithmic trading agents exploit. When agents develop price-altering algorithms, any action taken to lower a price by one agent may be instantaneously matched by another.

Some may argue that this reflects an efficient market; however, if the shared strategy of price fixing is mutual knowledge, then the algorithms may support artificially and tacitly agreed-upon higher prices simply by not lowering them. Designing algorithms specifically to collude is not necessary for collusion to occur. AI plays an increasingly significant role in all decision-making. Through trial and error, algorithms can reach the same outcome. The nonexistence of intentionality and the fleeting decision span of the interaction raise concerns about monitoring.

The issue of monitoring also pertains to overseeing a system of systems. The ability to monitor market manipulation is affected by the fact that its impacts can have a butterfly effect, with small fluctuations leading to substantial consequences across systems, and spreading with the speed of a viral contagion.

The rise of the “intangible economy”- where assets such as data, software, knowledge embodied in patents, and related intellectual property seem increasingly important to economic success, marks a shift from developed economies moving toward knowledge-based models to a new revolution toward digital economies.

Fintech lenders currently serve more creditworthy borrowers in comparison to shadow banks, but they charge a higher interest rate, which is consistent with the notion that consumers are willing to pay a premium for a better user experience. Fintech firms have enhanced the efficiency of financial intermediation in credit markets.

If AI is not steered equitably and sustainably, it can exacerbate social problems, erode human autonomy, and amplify issues from our plantation past, including inherited inequality and intergenerational immobility. It can also lead to an unfair allocation of wealth, and contribute to a culture of distraction – that of “panem et digital circenses”. In other words, if we provide them with digital platforms to chat, make short video clips, and play games, they will forget to take action.