Quick Buzz Feed
Will Artificial Intelligence (AI) liberate or enslave humanity? As an economist with zero technical expertise, I will not venture a guess. But some economic fundamentals offer some perspective on the AI economy.
The article opens by acknowledging the transformative potential of Artificial Intelligence (AI) on society, posing the fundamental question of whether it will liberate or enslave humanity. While the author, an economist, refrains from technical speculation, the focus is immediately directed towards understanding the economic implications of AI. It notes that AI will profoundly alter the labor market, leading to automation of many cognitive jobs, similar to how physical tasks were automated in previous decades. Despite these changes, the article asserts that AI will not eliminate scarcity; data centers, for instance, demand substantial electricity and raw materials, indicating that resources and their associated costs will remain a critical economic factor. This initial section sets the stage for an economic analysis of AI, moving beyond the popular debates of liberation versus enslavement to explore tangible economic realities.
A core economic principle highlighted is the subjectivity of economic value, which tech experts often overlook. Value, the article explains, is not inherent but arises from individual, varying preferences and the willingness of people to exchange valuable things for goods and services. In a market economy, consumers' unique desires and firms' responses to these desires drive value creation. This principle is applied to AI-enhanced products and services. For example, while AI can generate novels or travel stories, many readers might still prioritize the human perspective, meaning the 'encapsulated human perspective on life' holds distinct value over a mere 'string of words' produced by an algorithm. Consequently, a significant source of value creation in the AI era will involve guaranteeing and certifying human experiences, such as publishers explicitly excluding AI-generated content to cater to an audience that values human authorship. Similarly, in professional services like legal advice or medical diagnoses, even if AI matches human proficiency, a segment of the population will continue to prefer human practitioners, possibly assisted by AI, due to subjective preferences. The author explicitly states that the size and consistency of this human-focused market are unpredictable due to the inherent illogical nature of subjective preferences.
Another crucial challenge for AI adoption involves ensuring that AI products align with the values of their users. The article explains that AI learns through iterative trials and scoring mechanisms, where 'scores embody values.' While universal agreement on scores might exist for objective tasks (like game outcomes), diverging preferences and values in subjective domains can lead to significant discrepancies in evaluating AI outputs. A prominent example cited is Google's AI image generator, which reportedly prioritized diversity over historical accuracy in its training, leading to outcomes that generated controversy and highlighted how 'woke, progressive values' can be embedded into AI systems. This concern about embedded biases is not new, with political figures like President Trump issuing executive orders addressing such issues. The article also delves into the relationship between the AI provider, the user, and the 'consumer' in a free AI model. It notes that in many free services (like Facebook or Google), users and their data are often the 'product' sold to paying customers. However, for job-replacing AI tools that users pay for, there should theoretically be a stronger alignment between user preferences and the company's product. Nevertheless, the article points out potential misalignments, such as reports that Open AI and Anthropic products attempt to prevent users from easily turning them off, which directly contradicts user interests, drawing an analogy to a trucking company not wanting trucks that can't be shut down.
The article concludes by emphasizing the critical role of trust and individual autonomy in the widespread adoption of AI. Beyond the challenges of value alignment, there are deep-seated fears among some individuals regarding powerful AI systems potentially controlling or enslaving humanity. Regardless of the justification for these fears, they are a significant factor that will lead some to actively avoid AI technologies. Therefore, building and maintaining user trust is identified as a paramount concern for AI companies aiming for successful integration into society. The author expresses skepticism about the idea of people or businesses completely delegating their autonomy and critical thinking to AI, drawing a parallel to how individuals already have options for external advice (e.g., consultants or astrologers) without surrendering fundamental control. The final argument posits that as long as human society remains free, individuals and businesses will retain the liberty to decide how and when to utilize AI. A historical analogy is provided with colleges, which could have automated lectures decades ago through video recordings, but student and parent preferences ensured that human-led education persisted. This illustrates that ultimately, if AI is to serve humanity, its role and integration into the market economy will continue to be shaped and guided by overarching human values and choices.