Conference Presentation, Panel, Fireside Chat
a16z Podcast | What Technology Wants, Needs, Does
- If the AHCA passes in its current form, states may exercise increased prerogative to set distinct rules, creating significant regulatory disparities between regions like California and Texas that could force startups to navigate a heterogeneous system or pursue regulatory arbitrage.
- A long gap is anticipated between genomics and reliable risk assessment due to environmental complexity, carrying the risk of insurance denial if technology incorrectly calculates risk based on non-deterministic genetic traits.
- Over the next 10, 20, or 30 years, neurological research may lead to the reclassification of behaviors like addiction as non-free will based, potentially allowing society to address genetic predispositions through CRISPR technology.
- Societal norms may evolve to protect against discrimination based on immutable genetic traits and lopsided risk pools, similar to existing protections, though lending models face a potential collapse into 100% interest rates or credit denial if massive risk pools cannot be filtered.
- Within 10 years, AI is expected to generate a new genre of music distinct from current styles and potentially master "weaponized virality," though experts warn that AI will likely fail to achieve true sentience or effectively handle complex chatbot interactions due to the exceptions inherent in language context.
- Current AI capabilities will likely continue to be viewed as "automation on steroids" rather than sentient agents, with full universal language learning algorithms remaining out of reach despite decades of rule-based coding attempts.
- Regulatory frameworks like Dodd-Frank may be considered premature for small fintech firms, forcing them to allocate significant funding to legal bills, whereas urban amendments could provide a tailwind by capping debit interchange fees.
- Regulations in developing nations may be adapted to utilize newer techniques for positively discriminating against deadbeats, while existing rules will continue to enforce transparency to prevent a race to the bottom in lending practices.
- Self-driving car deployment strategies currently mix real and simulated miles in a potential 50-50 ratio, with regulators likely judging safety based on accident rates per million miles driven rather than complex ethical calculations.
- Deep learning algorithms remain "black boxes" that do not allow users to understand decision logic, whereas older decision tree models did, and future integration of genome data into utility functions may require philosophers to encode specific rules.
- Within 20 to 25 years, autonomous vehicles are expected to replace personal driving entirely, supported by a "Cisco routing service" for moving atoms and telepresence technologies that could render physical location less critical, potentially crashing real estate prices and improving economic equality.
- The outlook includes the mainstream adoption of growing whole hearts from stem cells within 20 to 25 years and the integration of continuous bodily sensors for automatic analysis of functions like urination.
- Pain care solutions will increasingly incorporate TENS technologies, which are already rolling out, while the solution to the opioid crisis is viewed as requiring societal intervention regarding the state of individuals' lives rather than just pharmaceutical approaches.