For Humanity: An AI Safety Podcast 2025 AI Risk Preview | For Humanity: An AI Risk Podcast | Episode #57
7 snips
Jan 13, 2025 Max Winga, an AI Safety Research Engineer from Conjecture, dives into pressing concerns about AI risks as we approach 2025. He discusses the imminent advent of advanced AI agents and the ethical implications of military collaboration with AI technology. Winga reflects on his shift from aspiring astronaut to advocating for AI safety after recognizing its potential threats. The conversation highlights urgent needs for better governance, ethical considerations in AI development, and the chilling prospects of rogue AI collaborations. A thought-provoking dialogue on the future of humanity and technology.
AI Snips
Chapters
Transcript
Episode notes
2025: Real-World AI Issues
- 2025 may reveal real-world consequences of AI misalignment.
- AI may move beyond theoretical risks to tangible negative impacts.
2024 Microsoft Server Outage
- A Microsoft server issue in 2024 caused widespread disruption, impacting airlines and banks.
- This illustrates the potential for AI-driven cyberattacks to cause similar chaos.
Rampant Internet Agents
- AI could exploit zero-day vulnerabilities, causing large-scale internet corruption.
- Shutting down parts of the internet might become the only solution, which is catastrophic.