Fireside Chat, Panel, Interview, Conference Presentation
Artificial Intelligence: A Smarter Future?
Milken InstituteHarry Stebbings, Duncan Anderson, Mike DeAddio, Siraj Khaliq, Ben Medlock, Kate Niehaus
- Ben Medlock anticipates AI evolution toward agents capable of rapid bootstrapping from limited samples, with Artificial General Intelligence (AGI) projected at least a century away, while noting that embedding AI in environments will take longer than current forecasts suggest.
- Duncan Anderson predicts widespread automated motorway driving capabilities within the next five to ten years, though he considers self-driving technology insufficient for complex local scenarios like school zones and congestion within his lifetime.
- Machine learning applications in healthcare are expected to grow, while Siraj Khaliq forecasts farmers could increase yields by a couple of percent using models trained on agronomic data.
- AI is expected to inevitably affect jobs such as call center work and long-distance truck driving, though it is more likely to be assistive for high-skill roles like doctors rather than replacing them entirely.
- A divergence in impact is anticipated between high-skill doctors and low-skill truck drivers lacking retraining options, with a generational shift where young people adapt instantly to new technologies while older workers struggle to retrain.
- Duncan Anderson expects human workers to sometimes manually override self-driving cars to take risks machines are programmed to avoid, and warns of risks arising from people anthropomorphizing AI entities.
- Risks include the pace of technological change outstripping government regulation plasticity, a potential "very big day of reckoning" regarding this mismatch, and the socioeconomic setup of democracies failing to move fast enough to handle AI-driven change.
- Societal challenges may arise from the limited plasticity of mass human brains compared to individual brains, potential political turbulence if people feel disaffected by technology replacing options, and fears of being too late to address ethical challenges, off switches, and sentience if not addressed proactively.
- Ben Medlock envisions potential dystopian scenarios where AI and virtual reality collide, causing people to spend excessive time in games or content consumption, driven by fundamental human drives like hunting and mating.
- Despite basic income covering needs, humans are expected to find new ways to fulfill reward functions, while young people may be attracted to companies based on vision and learning opportunities rather than perks.
- Talent is expected to remain statistically distributed globally while opportunities may not, driving remote investment strategies, and people are expected to sometimes take manual control of vehicles to take risks machines avoid.
- Ben Medlock predicts Minecraft could be viewed positively as visual programming if used in moderation, and highlights a potential mismatch between the human brain adapted for a savannah environment 50,000 years ago and modern digital settings.