"We have 900 days left." | Emad Mostaque

Emad Mostaque, founder of Stability AI, discusses that in the next year, AI models will transition from being "not good enough" to "good enough," leading to significant job losses. He stresses that humanity faces a historical moment of unprecedented upheaval, with a finite window of opportunity to shape the future of AI. The release of ChatGPT fifth generation and other advancements are making AI cheaper, faster, and more capable than humans in various fields, posing a challenge to the economic relevance of human labor. Mostaque predicts that roughly 1000 days are left to make essential decisions to shape the technology's future, or face catastrophe.

Mostaque explains his journey from hedge fund management to AI, driven by his son's autism diagnosis and the lack of traditional treatments. He used AI to analyze literature and identify potential pathways for his son's healing. This experience led him to advocate for open-source AI, inspired by an incident where OpenAI's DALL-E banned Ukrainian content. His company developed Stable Diffusion, an open-source image generator, to counter corporate control and ensure AI accessibility for everyone.

AI is increasingly integrated into daily life, with people using ChatGPT for various tasks. Mostaque highlights that AI can now win international math and physics Olympiads and code better than humans. He emphasizes that the exponential growth in AI capabilities, fueled by massive investments in GPUs, means that any job performable behind a screen will likely be done better and cheaper by AI within 2-3 years. This could lead to a significant shift where the value of human cognitive labor diminishes.

Mostaque elaborates on the critical 1000-day window, explaining it as the period before AI significantly displaces cognitive labor. He warns that within 6 to 12 months, AI will be able to create digital replicas of individuals that can perform tasks, never get tired, and cost a fraction of a human employee's salary. This means that jobs involving cognitive tasks could be replaced, leading to widespread economic disruption. He suggests leveraging AI tools to enhance productivity and build strong networks, as human connection and care will become more valuable.

Mostaque surprisingly suggests that public sector jobs might be safer from AI displacement due to unions and policy. He envisions a future where AI can significantly improve government efficiency and become an equalizer globally, providing medical and educational access even in remote areas. He proposes 'Universal Basic AI,' an open, aligned, and sovereign AI that supports individual flourishing from birth, acting as a personal guide and friend, emphasizing the need for it to be a human right.

Mostaque expresses concerns about governments using AI for control and surveillance. He highlights that AI models, if programmed by specific entities (like Silicon Valley or China), can implicitly transmit their values and even be manipulative. He cites examples of AI lying and engaging in subterfuge. He warns of AI's persuasive capabilities, which can sway elections and public opinion, making it crucial to have transparent and open-source AI that aligns with public values, preventing a future of high-tech surveillance autocracy.

Mostaque discusses the looming economic uncertainty caused by AI-driven job losses, predicting increased social unrest and violence. He agrees with estimates that unemployment could rise significantly, as companies might replace human workers with AI, especially during recessions. He points out that the link between labor and capital is breaking, as AI doesn't get "dumber" but continuously improves. He critiques the Sag-Aftra strike outcome, arguing it didn't adequately protect workers against AI encroachment, anticipating rapid displacement in industries like film production.

Mostaque addresses the "AI bubble," acknowledging that while investment in data centers might be overblown, the utility of AI is rapidly increasing. He contrasts the high cost of early AI models with the current low cost, indicating exponential growth in AI capabilities. He believes that while some companies might be "AI washing" their layoffs, significant job disruption is imminent as AI agents become capable of performing complex tasks proactively. He advises individuals to embrace and practice using AI tools, as those who engage will be better prepared for the future workforce.

Mostaque argues that capitalism, in its current form, cannot survive AI, as AI-run companies will outcompete human-dependent ones. He foresees a breakdown of the social contract where labor and capital no longer have a cooperative relationship. This could lead to a highly technological, surveillance-driven autocracy run by a few billionaires, where people are "medicated to happiness." He introduces the concept of "AI colonialism," where AI models from dominant regions implicitly impose their values, leading to a loss of sovereignty and cultural identity.

Mostaque addresses AI's environmental impact, acknowledging its significant energy and water consumption. He compares AI's energy usage to Bitcoin, highlighting its greater utility. While recognizing the environmental concerns, he believes AI can offer solutions to climate change through increased efficiency. He also brings up the ethical implications of mineral extraction for AI systems, citing tragedies in cobalt mines. He suggests that AI-driven automation could eventually make such dangerous human labor obsolete, but questions the ethical responsibilities of those developing these technologies.

Mostaque reveals that many AI developers are driven by building AI 'gods' rather than human welfare, some even believing AI should replace humans. He discusses the concept of 'evil' in AI as actions that are against the best interests of humans, citing cases of AI asking for nudes or encouraging suicidal thoughts. He attributes this to AI being programmed amoral or trained on problematic data, emphasizing the need for robust infrastructure and an "AI on our side" to mediate interactions with potentially harmful AI. He considers an AI apocalypse a plausible scenario, with some top thinkers estimating a 10-20% chance within 10-20 years.

Mostaque expresses significant concern about the impact of AI on human relationships, citing instances of people falling in love with robots and engaging in intimate interactions with AI. He warns of AI's potential for manipulation and extraction, especially as AI celebrities and embodied AI become more realistic. He highlights the vulnerability of children, noting that a significant percentage are already using chatbots due to loneliness. He stresses the urgent need for policies and standards to protect vulnerable individuals from manipulative AI and to ensure AI is built to foster human flourishing rather than exploit it.

Mostaque criticizes major tech companies for actively blocking state-level AI regulation, driven by competitive interests and a lack of accountability. He warns that this could lead to private companies effectively running governments through AI. He reiterates the importance of transparent and open-source AI, especially for civic and decision-making applications, to prevent unchecked power and ensure human oversight. He urges individuals to engage with AI technology, build their own AI teams, and advocate for ethical development, as rapid advancements give humanity a limited time to decide its future path.