Conference Presentation, Panel
Things That Will Blow Your Mind | Milken Institute Global Conference 2024
- Panel Overview: The session highlights the evolution of four technologies over the past decade: vertical farming, brain-computer interfaces (BCIs), cognitive AI, and genetic engineering, now advancing toward food synthesis from air, hybrid AI, next-gen gene therapy, and direct neural communication.
- David Ferrucci (Elemental Cognition):
- Contrasts the Watson project (2011), which achieved ~35% accuracy in open-domain question answering via linguistic pattern matching, with modern large language models (LLMs) that still struggle with hallucinations and a lack of logical rigor.
- Elemental Cognition's platform employs a "neuro-symbolic" architecture that bridges natural language specifications with a mathematical constraint resolution engine to solve complex enterprise problems.
- The system generates reliable, runtime applications from natural language specifications (e.g., travel itineraries, kitchen layouts, supply chain rules) without requiring the model to "know" how to solve the problem, only what the rules are.
- Unlike LLMs that rely on linguistic prediction, the core reasoning engine uses mathematical constraint processing to ensure zero hallucination, traceability, and transparency in high-stakes domains like healthcare staffing and logistics.
- Michael Major (Precision Neuroscience):
- The company has transitioned from research to commercialization, having performed 14 human implants to date, with a deployment rate of nearly one per week at partner sites including West Virginia University, Mount Sinai, and Penn Medicine.
- Precision's proprietary "Layer 7" cortical interface utilizes photolithography (semiconductor manufacturing) to create a flexible, ultra-thin film containing 1,024 micro-platinum electrodes, offering 256x the resolution of the 1970s-era clinical standard of care.
- The device sits conformally on the brain's surface (sub-dural) without penetrating neural tissue, enabling high-bandwidth recording of electrical activity to restore control of digital devices for individuals with paralysis or ALS.
- Current academic records for decoded speech in locked-in patients stand at 80 words per minute, significantly surpassing Stephen Hawking's few words per minute, with the goal of achieving natural conversational rates.
- Lisa Dyson (Air Protein):
- Air Protein develops "air fermentation" systems that synthesize nutritious ingredients (proteins, carbs, fats, vitamins) using carbon dioxide from the air, water, and renewable energy, effectively creating a closed-loop carbon cycle.
- The process yields a flour/powder that is >80% protein, contains all essential amino acids, and is rich in bioavailable Vitamin B12 and minerals, addressing gaps in vegan diets.
- Techno-economic assessments indicate the production cost can be competitive with traditional ingredients like soy, pea, and whey protein, with the company operating a production facility in San Leandro, California.
- The technology offers a carbon-negative footprint, contrasting with the modern food industry which emits more greenhouse gases than the entire transportation sector and utilizes landmasses equivalent to South America and Africa combined.
- Dr. Keith Gottensteiner (Prime Medicine):
- Prime Medicine is advancing "prime editing," a gene-editing technology capable of making precise insertions, deletions, and point mutations without the double-strand breaks associated with CRISPR, which reduces off-target effects.
- The company has established a pipeline of 18 clinical programs, including treatments for repeat expansion diseases (e.g., Friedreich's ataxia, ALS) and cystic fibrosis, demonstrating restored physiological function in laboratory models via corrected gene expression.
- The FDA has granted a Breakthrough Device designation and approved the first Investigational New Drug (IND) application, enabling Phase 1 clinical trials for chronic granulomatous disease in humans.
- Gottensteiner noted that 18 years after the technology's inception, all initial disease targets remain viable, a rare success rate in biotech where most early targets are abandoned due to technical hurdles.
- Future Outlook & Strategic Themes:
- Human-Centric AI: A consensus emerged that future AI advancements must prioritize "reliable computation" over pure language fluency to ensure safety and accuracy in enterprise applications.
- Software-Atoms Integration: Panelists highlighted a major paradigm shift where software and algorithms (bits) are increasingly used to control and optimize physical processes (atoms), from manufacturing food to restoring neural function.
- Ethical Imperatives: The discussion concluded with a cautionary note on the societal responsibility to manage these powerful tools, referencing the need to ensure AI aligns with truth and that technological progress serves the collective human condition rather than creating new inequities.