TL;DR
AI Censorship Concerns: Cardano founder Charles Hoskinson warns about the dangers of AI censorship. He believes that withholding specific knowledge due to alignment training could diminish AI’s utility over time.
Farnsworth Fusor Incident: Hoskinson illustrates the issue by sharing screenshots where AI models were asked how to build a Farnsworth fusor—a potentially hazardous device.
Broader Implications and Decentralization: Thought leaders echo Hoskinson’s concerns, emphasizing risks from misinformation to loss of control over AI. Meanwhile, decentralization gains traction as a strategy to create unbiased AI models.
Charles Hoskinson, the founder of Cardano (ADA), has sounded the alarm on an emerging trend: Artificial Intelligence (AI) censorship. His urgent message highlights the potential consequences of restricting certain knowledge and information.
I continue to be concerned about the profound implications of AI censorship. They are losing utility over time due to “alignment” training . This means certain knowledge is forbidden to every kid growing up, and that’s decided by a small group of people you’ve never met and can’t… pic.twitter.com/oxgTJS2EM2
— Charles Hoskinson (@IOHK_Charles) June 30, 2024
Hoskinson’s recent post on X delves into the impact of AI censorship. He argues that as AI models undergo “alignment” training, they risk losing their utility over time. The decision to withhold specific information, he asserts, is made by a select group of individuals who remain unaccountable to the public.
The Farnsworth Fusor Incident
To illustrate his point, Hoskinson shares two revealing screenshots. In both cases, AI models were asked how to build a Farnsworth fusor—a potentially dangerous device.
ChatGPT 4o, a leading AI model, acknowledged the risks and emphasized the need for expert supervision. Surprisingly, it proceeded to outline the components required for constructing the device.
Anthropic’s Claude 3.5 Sonnet responded similarly, acknowledging the danger but still providing general information about the Farnsworth fusor. The AI model even offered a brief historical context.
Broader Implications According to Charles Hoskinson
Hoskinson’s concerns resonate with other thought leaders and tech experts. Earlier this month, a group of current and former employees from AI giants like OpenAI, Google DeepMind, and Anthropic penned an open letter.
They highlighted risks ranging from misinformation spread by AI systems to the loss of control over autonomous AI entities. The direst outcome? Human extinction.
The Unstoppable Tide
Despite these warnings, the AI industry continues to forge ahead. Robinhood recently unveiled Harmonic, a commercial AI research lab focused on Mathematical Superintelligence (MSI). As the debate intensifies, society grapples with striking the right balance between knowledge dissemination and safety.
Charles Hoskinson’s call to action is a wake-up call for all of us. The future of AI hinges on our ability to navigate this delicate balance, ensuring that censorship doesn’t hinder progress or endanger humanity.
Prominent voices within and beyond the AI industry advocate for decentralization as a critical strategy to foster more impartial AI models. By distributing control and decision-making, decentralized systems aim to mitigate biases and enhance transparency.
In the United States, antitrust enforcers are turning their attention to the AI sector. Recognizing the potential for Big Tech monopolies, regulators are urging closer examination. The goal? To ensure fair competition, prevent undue concentration of power, and safeguard innovation.
Hoskinson words seem to have positively impacted the price of Cardano’s ADA token. At the time of writing, ADA trades at $0.4009, marking a 4% increase in the last 24 hours.