AI Explained: The BonsXAI Blog
Liberty in the Age of AI: Beyond Voluntary Commitments
Author: Peter Ball
Last week I wrote about how corporate shareholder maximization was "our Skynet moment" sucking humanity into a doom loop based on the 'greed is good' meme. Unchecked AI proliferation threatens to accelerate, scale and obfuscate this vicious cycle.
As Yogi Berra said, "It's tough to make predictions, especially about the future". Nonetheless, "The Fourth Industrial Revolution" predicts a dramatically different future as new economic ecosystems arise from rapidly increasing interconnectivity and automation. As modern technologies such as AI, web3, IoT, and quantum computation advance, their convergence will lead to a variety of "More is Different" (see the 1972 Science article by P. W. Anderson) economic and social phenomena. Imagine trillions of sensors feeding data to billions of autonomous agents transacting business on cryptography-based distributed ledgers day and night driven by AI systems developed, at least in part, to generate competitive returns on investment. As these agents transact, they change the very ecosystems from which their input data is collected, forming a dynamic tangle of interactive feedback loops. Yes, it's tough to make predictions, especially when you breach "More is Different" thresholds (think 'phase transitions'). Because "More is Different", we cannot know what flaws may be embodied in these systems, but whatever flaws do arise could potentially be scaled worldwide in days, hours or minutes. Will someone suffer the consequences of unforeseen and unmitigated risk? Will anyone even know?
The White House recently announced that a number of tech companies including Google, Microsoft, Amazon, Meta, OpenAI, Anthropic, and Inflection have voluntarily pledged to a number of commitments to develop safe AI technology. The specific commitments include:
- Developing ways to identify AI-generated content.
- Engage independent experts to asses AI tools before public release.
- Share information on best practices.
- Allow third parties to identify vulnerabilities in AI systems.
- Disclose limitations and guidance on appropriate use of AI systems.
Given that each of these tech companies arguably plays the role of Skynet overlord, how much faith should we have in their voluntary commitments? While pledges make for good public relations, these big tech companies still have a glaring fundamental conflict of interest. So long as they are obliged or incentivized to increase shareholder value, they'll be foxes guarding the henhouse.
The word "liberty" is often defined as a state of being free from oppressive restrictions imposed by authorities. As AI systems advance, their apparent and actual authority becomes ever more established. But what about freedom from ignorance, misinformation and propaganda? A broader definition of liberty includes freedom from abusive or hidden manipulation. To protect this broader version of liberty, voluntary commitments are inadequate. Effective AI regulations are essential.
Around the world, policy makers, governments, and academics are developing a variety of governance systems and regulatory regimes designed to keep people safe from unchecked AI systems. At bonsXAI, we're following and studying these regulatory efforts and building a comprehensive suite of tools to enable you to meet the challenges of AI compliance. In my next post, I'll start investigating these efforts.
