AI and Minshall’s Man Crab

Soon, AI will play an increasingly essential role in criminal acts. Certain crimes would not occur “but for” the use of AI. “Only a god can save us.” So said Martin Heidegger, in an interview with Spiegel on September 23, 1966. Humanity is at an unprecedented crossroad. Difficult decisions must be made about how we are to live within the infosphere. With AI’s advance, our only possibility of salvation is in readiness of expectation.

Peter Minshall reasons that many of us hold the view that all men are created equal in the eyes of God, but not one of us has braved the question – “Are all Gods Equal in the Eyes of Men?” Heidegger and Minshall light pitch oil lamps beneath the yellow sun in broad daylight. Minshall’s Man Crab searches technological society to spotlight the danger AI poses to civilization. Heidegger on the other hand helps us to understand that reliance on technology and information alone cannot help us avert the x-risks of AI.

AI is a growing reservoir of interactive, autonomous, and self-learning agency that can deal with tasks successfully independent of the creators of such Artificial Agents. This combination of learning, skills, and autonomy underpins both beneficial and malicious uses of AI. Social media is the modern Woodford Square. To avoid replicating old hierarchies that reinforce unequal distributions of social, economic, cultural, and political power, governments must commit to designing AI for democracy.

In the 2017 case Packingham v. North Carolina, the Supreme Court declared that the Internet is “the modern public square.” However, instances of what may be determined to be harassment can become enmeshed with the use of AI Deepfake Digital Clones to exercise free speech. Malevolent actors can deploy a social bot as an instrument of direct or indirect harassment. Direct harassment is constituted by propagating and spreading hateful messages against a person. Indirect approaches include retweeting or liking negative posts and skewing polls to give a fake impression of wide-scale bitterness against a person.

Proponents of free speech point out that the digital public square is the quintessential site of democratic deliberation and civic participation, a physical “marketplace of ideas.” The doctrinal and policy consequences that flow from the analogy are extraordinary. The digital-public-square view chiefly emphasizes the principle of openness. Accordingly, adherents to this perspective tend to view restrictions and regulations as antidemocratic and censorious.

Five types of criminal offenses that can be potentially affected by AI include (1) commerce, financial markets, and insolvency (including trading, and bankruptcy); (2) harmful and dangerous drugs (including illicit goods); (3) offenses against the person (including homicide, murder, manslaughter, harassment, stalking, and torture) (4) sexual offenses (including rape, sexual assault); and (5) theft and fraud, and forgery and personation. The reasons for concern in these areas touch on issues surrounding: (1) the autonomous emergence of coordinated actions, (2) the undermining of existing models of liability, (3) monitoring attribution, feasibility, and cross-system actions, and (4) human-bot interaction.

The use of AI for interrogation and torture is motivated by its capacity for affect-modelling. While torture is an ineffective method of extracting information, malicious actors may perceive the use of AI as a way to optimize the balance between misery and causing the subject to become confused and unresponsive. When this happens independent of human intervention, the mere distancing of a threat actor from the “actus reus” makes the use of AI during torture a unique menace.

Crimes that focus on the economy can include cartel offenses, such as price fixing and collusion, trading securities based on private business information, and market manipulation. Problems arise particularly when AI is involved in collusion, price fixing, and market manipulation. Simulation-based models of markets using artificial trading agents have shown that, through reinforcement learning, such agents can learn the techniques involved in placing orders with no intention of ever executing them.

Near-instantaneous pricing information is achieved when agents design price-altering algorithms. Any action to lower a price by one agent may be matched instantaneously by another. The absence of intentionality, the brief decision span, and the likelihood that collusion may emerge because of interactions among artificial agents raises serious concerns about liability and monitoring.

In the case of financial crime involving AI, monitoring is difficult because of the speed and adaptation of artificial agents. Artificial trading agents adapt and alter our perception of markets. Furthermore, the ability of AIs to learn and refine their capabilities implies that these agents may evolve new strategies, making it very difficult to detect their actions.

Social bots could exploit the cost-effective scaling of conversational and one-to-one advertising tools with hundreds of thousands of people every day to facilitate the sale of illegal drugs. Criminals can also subvert another actor’s buddy bot by skewing its learned classification and generation data structures using conversation.

Another worry is the growing practice of using data points from data sets across sectors to improve decision-making, collaboration, accountability, and transparency among experts who work in the humanitarian sector, and on human rights issues. It is also the case that issues related to fairness, accountability, or transparency (FAccT) in and around human rights data sharing can produce “ironic” consequences.