Keep pulling the thread on Eric Larsen.
Eric Larsen asserts that each new skill or plugin release from large language models like Claude can eliminate hundreds of billions of dollars in market capitalization from established industries.
Eric Larsen estimates that adversarial friction between payers and providers in the US healthcare system accounts for one trillion dollars in costs.
Eric Larsen predicts that AI-designed molecules and biologics will eventually reach the market at 1/20th the cost and 20 times the speed of traditional methods.
Eric Larsen predicts that healthcare startups, including highly valued decacorns, have a limited window to integrate into clinical workflows before the generalized capabilities of large "God models" surpass them.
Eric Larsen predicts that one of the hyperscale cloud providers, such as Microsoft, Google, or Amazon, will ultimately assume the product and medical malpractice liability for clinical AI.
Eric Larsen predicts that net profits for health insurance underwriting will trend asymptotically towards zero.
Eric Larsen predicts that AI's long-term impact on jobs will be mostly substitutive, contradicting the historical trend of technology creating more jobs than it destroys.
One out of every seven adult laborers in the United States works in the healthcare industry.
The US healthcare industry employs 23.8 million workers.
Eric Larsen believes that incumbent healthcare organizations are largely stuck in "point proof of concept purgatory," with very few AI pilots scaling to production.
Eric Larsen believes the US healthcare industry has been insulated from AI-driven disruption due to its regulatory enclosure and societal concerns about moving too quickly.
According to Eric Larsen, citing a piece by David Brailer in Health Affairs, the initial phase of AI adoption in healthcare is inflationary because new AI-driven models are operating alongside the legacy reimbursement system.