Karen Hao on AI's 'Imperial' Ambitions

Karen Hao, author of the international bestseller 'Empire of AI: Inside the Reckless Race for Total Domination,' joined Mishal Husain on The Mishal Husain Show to discuss the consolidation of power within the top AI companies. Hao argues that these companies, much like historical empires, are accumulating significant economic and political influence through the extraction of resources, data, and labor.

The 'Imperial' Nature of AI Companies

Hao draws a stark parallel between modern AI giants and historical empires, noting that while they may not engage in overt violence, they "seize and extract precious resources to feed their vision of artificial intelligence." This process, she explains, involves the "dispossession of resources, labor, value from the majority." She highlights how these companies claim ownership of individual data and intellectual property from artists and creators, and exploit labor through both the direct workers involved in producing their technologies and the workers whose jobs are automated away.

Exploitation in Data Work

The conversation delved into the often-overlooked "data work" category, which includes tasks like content moderation and data labeling. Hao points out that this is now the fourth fastest-growing job sector in the U.S., attracting highly educated individuals like scientists and lawyers. However, these workers often face "deeply exploitative, poorly paid conditions," working in a gig economy model where they are on call and pitted against each other. Hao has observed these patterns of exploitation for years, noting their consistency across different communities.

The full discussion can be found on Bloomberg Podcast's YouTube channel.

The Problem with Focusing on Leadership

Addressing the tendency to focus on the personalities at the top of AI companies, Hao asserts that the fundamental issue lies in the governance structure, not just the individuals. "We shouldn't have to live with companies that are so vastly powerful where one person at the top is able to decide all these things," she stated. She criticizes the obsession with identifying "good guys" and "bad guys," arguing that this distracts from the systemic problems of concentrated power and lack of accountability.

A Corrosive Approach to AI Development

Hao critiques the prevailing approach to AI development as a "reckless race for global domination." She questions why companies are not held to higher standards, especially when they can achieve their goals without such "corrosive" methods. She uses the analogy of transportation, comparing resource-intensive AI models to rockets, which are only deployed for specific needs, rather than ubiquitous tools like bicycles. Hao advocates for a shift towards more specialized and efficient AI technologies that have a less detrimental impact.

The Case of Malta and Data Harvesting

The discussion touched upon the agreement between Malta and OpenAI, where residents are offered free access to ChatGPT Plus and AI literacy courses. While governments may see this as beneficial, Hao argues that these countries are being exploited. "They are having their data harvested to train the next generations of models," she explained, emphasizing that these partnerships are driven by the companies' insatiable need for data, especially as high-quality internet data becomes scarce.

Power Dynamics: AI vs. Governments

Hao believes that AI companies are increasingly wielding more power than governments, even in the U.S. She cited the recent spat between Anthropic and the Department of War, where Anthropic, despite threats, did not fall in line, demonstrating the companies' willingness to push back against governmental authority due to their significant benefits and profits elsewhere. This, she argues, highlights a critical governance gap.

Alternative AI Models and a Call to Action

Hao champions the development of specialized AI and points to DeepMind's AlphaFold as an example of a "bicycle of AI", a technology that is resource-efficient, requires little data, and avoids labor exploitation while providing significant benefits. She stresses that there are many ways to produce AI without a "corrosive approach." Ultimately, she calls for leaders in all sectors to prioritize human agency and dignity in the face of AI's transformative potential.

Personal Philosophy and Future Work

Hao, who trained as a mechanical engineer before becoming a journalist, stated that her mission is to ensure everyone can live a dignified life. While she is currently focused on holding AI companies accountable, she hopes this will not be a lifelong fixation. She categorically stated she would never work for the "empires of AI" but would consider companies developing AI with a focus on human agency and sustainability. She cited Te Hiku Media, a non-profit that developed an AI speech recognition tool for the Māori language with community consent, as an example of a different, more ethical approach.

Conclusion: The Importance of Scrutiny

Hao concluded by stating she uses neither ChatGPT nor Claude, opting for alternatives that she feels liberate her. Her perspective underscores the critical need for public scrutiny and accountability in the development and deployment of AI, moving beyond market forces and founders' visions to ensure a future that benefits humanity.