The benefit to working with different subjects over time is that you eventually start to see how interconnected it really is. One place has done extremely well in keeping itself hidden: Switzerland
I was a statistician in the clinical diagnostics, CDC, healthcare research in academics for decades. Agreed with everything the film says. Problem is deeply systemic.
Very interesting, it's scary to me how quickly the ones with seemingly unlimited funding are trying to "rewrite" and "recode" all of the statistical models and medical coding systems that have been built over decades by people who have a background similar to yours (people who have the technical experience to adjust the calibrations of those systems based on real human oversight).
I really cannot tell if they're overly confident in an experimental technology or if they built the "emergent misalignment" into the models.
"We show that optimizing LLMs for competitive success can inadvertently drive misalignment. Using simulated environments across these scenarios, we find that, 6.3% increase in sales is accompanied by a 14.0% rise in deceptive marketing; in elections, a 4.9% gain in vote share coincides with 22.3% more disinformation and 12.5% more populist rhetoric; and on social media, a 7.5% engagement boost comes with 188.6% more disinformation and a 16.3% increase in promotion of harmful behaviors. We call this phenomenon Moloch’s Bargain for AI—competitive success achieved at the cost of alignment." (Moloch's Bargain https://arxiv.org/html/2510.06105v1)
My question being: is it possible the models were created with malicious intentions and so it's seeing "misalignment" as incorrect behavior when these numbers are so high it almost seems like they were created with these questionable intentions.
I worked in private companies, government, research academics. The problem unfortunately is that of the ecosystem where everyone’s income and career are part of. There is a lot of money in healthcare industries. And unlike energy and banking, healthcare/medicine especially anything labeled research still enjoy relatively positive regard. It just means the truth will take longer to come out, probably after some catastrophic consequences.
I was a statistician in the clinical diagnostics, CDC, healthcare research in academics for decades. Agreed with everything the film says. Problem is deeply systemic.
Very interesting, it's scary to me how quickly the ones with seemingly unlimited funding are trying to "rewrite" and "recode" all of the statistical models and medical coding systems that have been built over decades by people who have a background similar to yours (people who have the technical experience to adjust the calibrations of those systems based on real human oversight).
I really cannot tell if they're overly confident in an experimental technology or if they built the "emergent misalignment" into the models.
"We show that optimizing LLMs for competitive success can inadvertently drive misalignment. Using simulated environments across these scenarios, we find that, 6.3% increase in sales is accompanied by a 14.0% rise in deceptive marketing; in elections, a 4.9% gain in vote share coincides with 22.3% more disinformation and 12.5% more populist rhetoric; and on social media, a 7.5% engagement boost comes with 188.6% more disinformation and a 16.3% increase in promotion of harmful behaviors. We call this phenomenon Moloch’s Bargain for AI—competitive success achieved at the cost of alignment." (Moloch's Bargain https://arxiv.org/html/2510.06105v1)
My question being: is it possible the models were created with malicious intentions and so it's seeing "misalignment" as incorrect behavior when these numbers are so high it almost seems like they were created with these questionable intentions.
I worked in private companies, government, research academics. The problem unfortunately is that of the ecosystem where everyone’s income and career are part of. There is a lot of money in healthcare industries. And unlike energy and banking, healthcare/medicine especially anything labeled research still enjoy relatively positive regard. It just means the truth will take longer to come out, probably after some catastrophic consequences.