As we transition from AI strategies focused on business applications and ascend toward the pinnacle of the pyramid, the socio-economic considerations surrounding AI become increasingly intricate. When contemplating the entirety of society, the economic aspects of AI no longer remain straightforward.
The emergence of AI brings forth numerous choices for society, each accompanied by its own set of compromises. In this phase, where technology is approaching maturity, there are three notable trade-offs that hold significant importance at the societal level.
Productivity vs. Distribution
Stephen Hawking wrote in 2016, “The automation of factories has already decimated jobs in traditional manufacturing, and the rise of artificial intelligence is likely to extend this job destruction deep into the middle classes, with only the most caring, creative or supervisory roles remaining.” Several studies have already tallied up potential job destruction due to automation, and this time it isn’t just physical labor but also cognitive functions previously believed immune to such forces. After all, horses fell behind on horsepower, not brainpower.
While many argue that AI will make us economically poorer or worse off, this perspective is not entirely accurate. Economists concur that technological advancement ultimately improves our well-being and enhances productivity, and AI is bound to unequivocally contribute to productivity gains. However, the issue lies not in wealth creation but in wealth distribution. AI has the potential to exacerbate income inequality for two primary reasons.
Firstly, as AI takes over certain tasks, it may intensify competition among humans for the remaining tasks, leading to decreased wages and a further decline in the proportion of income earned by capital owners.
Secondly, prediction machines, along with other computer-related technologies, might exhibit a skill bias, disproportionately enhancing the productivity of highly skilled workers. This disparity in productivity gains can widen the income gap, contributing to income inequality.
Innovation vs Competition
Throughout history, technology has served as a fertile ground for the emergence of dominant companies. For instance, AT&T controlled the telecommunications industry for over five decades, while Microsoft and Intel held a monopoly in the field of information technology during the 1990s and 2000s. More recently, Google has dominated the search market, and Facebook has become the ruler of social media. These companies achieved significant growth by leveraging their core technologies, which allowed them to scale and attain lower costs and higher quality. However, it is important to note that technology-based monopolies are not permanent, as economist Joseph Schumpeter described in his term "creative destruction."
Similar to most software-related technologies, AI exhibits scale economies. Moreover, AI tools often exhibit increasing returns, wherein improved prediction accuracy attracts more users, resulting in the generation of more data, which in turn leads to further enhanced prediction accuracy. Businesses are incentivized to develop prediction machines when they have greater control over them, but this can also lead to monopolization alongside scale economies. While rapid innovation may initially benefit society in the short term, it may not be the optimal approach from a social or long-term perspective.
Performance vs Privacy
The third trade-off revolves around the balance between performance and privacy. AI systems demonstrate improved performance when they have access to more data, especially personal data, which enables them to personalize their predictions effectively. However, the provision of personal data often comes at the cost of reduced privacy. Certain jurisdictions, such as Europe, have opted to establish an environment that prioritizes privacy for their citizens. This approach can benefit individuals and potentially foster a dynamic market where people can make informed decisions regarding trading, selling, or donating their private data.
Conversely, this emphasis on privacy can create challenges in situations where the cost of opting in is high. It may also put European firms and citizens at a disadvantage in markets where AI systems with better access to data hold a competitive edge.
In contrast, China holds a significant advantage in the race for AI development due to three key factors. Firstly, China's substantial investment in AI, including significant projects, startups, and fundamental research, contributes to its unprecedented advantage. Secondly, China possesses an immense quantum of data as a result of its large population and widespread smartphone usage, offering a substantial resource for AI development. Thirdly, China's comparatively lower privacy protection measures provide both government and private sector entities with a notable advantage in optimizing the performance of their AI systems.
When considering the three trade-offs discussed, jurisdictions must carefully evaluate and balance both aspects of each trade-off. They need to assess the benefits and drawbacks associated with each choice and formulate policies that align with their broader strategies and the preferences of their citizens.
In the trade-off between AI and job displacement, jurisdictions need to consider the potential economic consequences of automation and AI adoption. They should assess the impact on employment and income inequality, while also recognizing the productivity gains that AI can bring. Designing policies that address these concerns, such as reskilling programs or social safety nets, can help mitigate the negative effects and ensure a more inclusive and equitable transition.
In the trade-off between technology-based monopolies and innovation, jurisdictions must weigh the advantages of faster innovation against the potential negative consequences of monopolization. They need to consider whether prioritizing short-term societal gains from rapid innovation is optimal or if a more balanced approach, accounting for long-term social well-being, is necessary.
Regarding the trade-off between AI performance and privacy, jurisdictions must take into account the benefits of improved AI performance through access to personal data. At the same time, they must address concerns related to privacy infringement and potential risks associated with data exploitation. Jurisdictions may choose to prioritize privacy protection, as seen in Europe, or adopt policies that strike a balance between AI performance and privacy concerns.
Ultimately, the decisions on how to navigate these trade-offs will depend on the specific goals, values, and priorities of each jurisdiction. By carefully considering both sides of the trade-offs and incorporating the preferences of their citizens, jurisdictions can design policies that strike a balance and align with their overall strategies for the adoption and regulation of AI technologies.