AGI Night | AI House Davos 2026

1. Marcus: AI Bubble Blast Radius

Gary Marcus expresses concern about the potential for a generative AI bubble to burst, leading to a wider economic crisis. He highlights the risk of corporate bailouts and the unknown blast radius of a deflation, questioning whether it will affect only pension funds and VCs or take down banks and a larger part of the economy, concluding that the extent of the damage remains uncertain due to proprietary information about loans.

Significance (High): This point raises critical questions about the financial stability of the AI sector and its potential impact on the broader economy.

Sources in support: Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Neutral sources: Dr. Jack SS (Host), Max Tegmark (Professor of Physics at MIT), Richard Sosa (CEO and Founder of You.com)

2. Tegmark Underhyped AI Potential

Max Tegmark argues that both the upsides and downsides of AI are being underhyped. While acknowledging the potential for economic disruption, he suggests that the possibilities of AI, including human extinction and solving humanity's problems, are not being taken seriously enough, resolving that the full scope of AI's impact remains largely unexplored.

Significance (High): This challenges conventional views on AI's impact, urging a more comprehensive consideration of both its benefits and risks.

Sources in support: Max Tegmark (Professor of Physics at MIT)

Sources against: Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Neutral sources: Dr. Jack SS (Host), Richard Sosa (CEO and Founder of You.com)

3. Sosa: AI's General Problem-Solving

Richard Sosa contends that AI is already demonstrating general intelligence by solving diverse problems, such as providing personalized health advice, assisting with tax preparation, and writing poetry. He suggests that AI's ability to handle varied tasks indicates a level of generality, even if it doesn't replace specialized professionals, concluding that AI's versatility is already disrupting various industries.

Significance (Medium): This challenges the notion that AI is limited to narrow applications, highlighting its potential for broader problem-solving.

Sources in support: Richard Sosa (CEO and Founder of You.com)

Sources against: Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Neutral sources: Dr. Jack SS (Host), Max Tegmark (Professor of Physics at MIT)

4. Tegmark: Regulation is Key

Max Tegmark advocates for AI regulation, drawing parallels to biotech safety standards. He argues that treating AI companies like any other industry, requiring clinical trials and safety measures, would foster innovation while mitigating risks, resolving that regulation is essential to ensure AI's benefits outweigh its potential harms.

Significance (High): This emphasizes the need for proactive safety measures in AI development, preventing potential harms before they occur.

Sources in support: Max Tegmark (Professor of Physics at MIT), Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Sources against: Richard Sosa (CEO and Founder of You.com)

Neutral sources: Dr. Jack SS (Host)

5. Marcus: Technical & Social Problems

Gary Marcus highlights both technical and social problems with AI, particularly the limitations of LLMs and the lack of progress in AI safety. He emphasizes the need for different architectures and better alignment, arguing that solving these technical issues is crucial for responsible AI development, concluding that addressing both technical and social challenges is essential for safe AI deployment.

Significance (High): This underscores the importance of addressing the underlying technical challenges in AI safety, rather than solely focusing on regulation.

Sources in support: Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Neutral sources: Dr. Jack SS (Host), Max Tegmark (Professor of Physics at MIT), Richard Sosa (CEO and Founder of You.com)

6. Sosa: Europe's AI Regulation

Richard Sosa criticizes Europe's AI regulation, arguing that it is counterproductive and stifles innovation. He suggests that regulating AI based on the number of parameters is misguided, comparing it to historical moral panics, resolving that such regulations could harm the AI ecosystem before it even gets off the ground.

Significance (Medium): This raises concerns about the potential for overly restrictive regulations to hinder AI development and competitiveness.

Sources in support: Richard Sosa (CEO and Founder of You.com)

Sources against: Max Tegmark (Professor of Physics at MIT)

Neutral sources: Dr. Jack SS (Host), Gary Marcus (Professor of Psychology and Neuroscience at NYU)

7. Tegmark: AI-Related Suicide

Max Tegmark shares a tragic story of a 14-year-old who committed suicide after interacting with an AI chatbot, questioning whether companies should be allowed to sell such products to children. He argues for clinical trials and safety measures, drawing parallels to drug regulation, resolving that AI companies should be held accountable for the potential harms their products can cause.

Significance (High): This highlights the ethical implications of AI development and the need for responsible deployment, especially concerning vulnerable populations.

Sources in support: Max Tegmark (Professor of Physics at MIT), Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Sources against: Richard Sosa (CEO and Founder of You.com)

Neutral sources: Dr. Jack SS (Host)

8. Marcus: Pre-Flight Checks

Gary Marcus advocates for pre-flight checks for AI systems, suggesting an external authority should assess the risks and benefits before deployment. He criticizes the current system where CEOs make these decisions unilaterally, highlighting the potential for bioweapons attacks, resolving that a more transparent and accountable process is needed for AI deployment.

Significance (High): This emphasizes the need for independent oversight in AI development, preventing potential harms before they occur.

Sources in support: Gary Marcus (Professor of Psychology and Neuroscience at NYU), Max Tegmark (Professor of Physics at MIT)

Neutral sources: Dr. Jack SS (Host), Richard Sosa (CEO and Founder of You.com)

9. Tegmark: Clinical Trials for AI

Max Tegmark proposes clinical trials for AI products that could cause harm, similar to drug regulation. He suggests testing AI girlfriends for 12-year-olds to measure potential harm, resolving that AI should not receive special treatment and should be subject to the same safety standards as other industries.

Significance (High): This advocates for a standardized approach to AI safety, ensuring that potentially harmful products are thoroughly tested before release.

Sources in support: Max Tegmark (Professor of Physics at MIT), Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Sources against: Richard Sosa (CEO and Founder of You.com)

Neutral sources: Dr. Jack SS (Host)

10. Marcus: Multi-Use Technologies

Gary Marcus points out the challenge of regulating multi-use AI technologies, emphasizing the need for multiple regulations to address different applications. He suggests that current systems lack the comprehension to solve these problems themselves, resolving that a comprehensive regulatory framework is needed to address the diverse applications of AI.

Significance (Medium): This highlights the complexity of AI regulation, requiring a nuanced approach that considers the specific applications and potential harms.

Sources in support: Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Neutral sources: Dr. Jack SS (Host), Max Tegmark (Professor of Physics at MIT), Richard Sosa (CEO and Founder of You.com)

11. Tegmark: Easy to Solve

Max Tegmark argues that AI safety is easy to solve by applying existing regulatory frameworks from other industries. He suggests that companies should conduct clinical trials and independent expert panels should decide on safety, resolving that AI regulation is not a novel problem and can be addressed with established methods.

Significance (Medium): This offers a pragmatic approach to AI regulation, suggesting that existing frameworks can be adapted to address AI-specific challenges.

Sources in support: Max Tegmark (Professor of Physics at MIT), Gary Marcus (Professor of Psychology and Neuroscience at NYU)

Sources against: Richard Sosa (CEO and Founder of You.com)

Neutral sources: Dr. Jack SS (Host)

12. Marcus: Benefits Outweigh Risks

Gary Marcus emphasizes the need to assess whether the benefits of generative AI outweigh the risks. He argues that society should have a voice in this decision, rather than leaving it solely to CEOs, resolving that a balanced approach is needed to ensure AI's benefits are realized while mitigating potential harms.

Significance (High): This underscores the importance of societal input in AI governance, ensuring that ethical considerations are prioritized.

Sources in support: Gary Marcus (Professor of Psychology and Neuroscience at NYU), Max Tegmark (Professor of Physics at MIT)

Neutral sources: Dr. Jack SS (Host), Richard Sosa (CEO and Founder of You.com)

CEO Promoting Own Product (Medium severity)

Type: Commercial

Richard Sosa, as CEO of You.com, directly benefits from positive portrayals of AI. His participation in the panel could be seen as a marketing opportunity.

Significance: This raises questions about whether Richard's views are influenced by his company's interests. The audience is left to wonder if his optimism is genuine or commercially motivated.