Artificial Intelligence: Threat to Human Activities

Welcome to learning edition of the Data Pragmatist, your dose of all things data science and AI.

📖 Estimated Reading Time: 5 minutes. Missed our previous editions?

• AI's Impact on Human Creativity

• Unsupervised Learning in Machine Learning

🧠Artificial Intelligence: Threat to Human Activities

The phenomenon of artificial intelligence (AI) is deeply embedded in our culture and widely utilized across various fields of human activity. Since its emergence, AI has captivated the attention of both scholars and the general public due to its almost mythological nature and its capabilities, which strikingly resemble those of living beings. This similarity and the unprecedented novelty of AI have spurred numerous speculations, including concerns about the threat AI might pose to humanity. These concerns range from purely fictional scenarios of a "rise of the machines" to serious assessments of risks in different fields, with the former being the most unrealistic yet paradoxically well-established in our consciousness.

Relevance of the Topic

• Ubiquity and Control: Understanding the relevance of the topic is crucial. The primary reason for this is the ubiquitous nature of AI. Machines capable of processing data are present in almost every aspect of our daily lives and in virtually any activity that requires data processing. This ubiquity leads to the assumption that AI's level of control is so overwhelming that humanity has almost relinquished its autonomy to another entity.

• Traits of Intelligence: This entity, AI, also exhibits traits previously thought to be exclusive to intelligent beings, often performing tasks more effectively than humans. Consequently, if AI ever "decides" to take control, there may be little humanity can do to oppose it. This narrative often highlights the rapid pace of AI development and its increasing capabilities, concluding that AI will inevitably become capable of making its own decisions, potentially disrupting the current state of affairs.

• Scientific Grounding: However, this debate rarely reaches the depth it deserves because these assumptions are often not scientific or evidence-based. Most AI specialists view the matter as an extension of popular fiction and dismiss it as unsupported by hard evidence. The common explanation involves the uneven distribution of AI capabilities: while current computers can perform basic mathematical functions millions of times faster than humans, they struggle with complex operations, particularly those requiring context interpretation, which humans routinely perform.

Artificial Superintelligence: Beyond Rhetoric

Despite the speculative nature of many AI-related fears, some concerns warrant serious consideration. Karamjit Gill, in his article "Artificial Superintelligence: Beyond Rhetoric" (2016), outlines several major issues that may pose real threats.

• Automation and Ethics: Gill points to advancements in automation technologies that require granting at least partial autonomy to AI for increased effectiveness. A pertinent example is the autonomous Google car, which will face ethical dilemmas such as prioritizing the life of a passenger versus the driver of an oncoming vehicle in a potential collision (Gill, 2016, p. 137). This issue extends beyond automated transportation, reflecting broader ethical challenges humanity has yet to resolve.

• Automated Weapon Systems: Gill also highlights the threat posed by automated weapon systems, which, rather than acting with malevolent intent, might make critical errors when handling multi-layered, context-sensitive data (Gill, 2016, p. 138).

• Economic Vulnerabilities: In the economic realm, Gill discusses the vulnerabilities introduced by AI. As the economy becomes increasingly digital, it becomes less transparent and harder for humans to control (Gill, 2016, p. 137). He further explores the concept of artificial general intelligence (AGI), a self-learning, self-regulating system that is challenging to monitor (Gill, 2016, p. 139). This obscurity leads to unpredictable outcomes and additional risks, difficult to estimate due to the novelty and rapid evolution of AI.

Conclusion

Gill's paper does not provide definitive answers but suggests directions for further inquiry, such as developing tools to manage current risks. Its value lies in distinguishing feasible and actual AI-related risks from those dismissed as unsubstantiated, thereby facilitating progress on addressing real threats. The paper underscores the importance of an ongoing dialogue between technology and society to navigate the complexities and risks posed by AI.

In summary, while fictional fears of malevolent AI can be dismissed, the potential adverse effects of AI on various aspects of human society warrant serious and continued examination.