no relation to the R package called “bayesmed”
Ahhh Jaan Tallinn, the Diet Coke of batfuckery. Jaan “Diet Batfuckery” Tallinn.
But will my insurance cover a visit to Dr. Spicy Autocomplete?
Oh hey a blog: https://www.encultured.ai/blog.html – Because of course they’re an rationalist AI alignment / AI gaming startup pivoting to healthcare
Part of their bold vision: AI agents that heal not just your cells, but also your society :')
Our vision for 2027 and beyond remains similar, namely, the development of artificial general healthcare: Technological processes capable of repairing damage to a diverse range of complex systems, including human cells, organs, individuals, and perhaps even groups of people. Why so general? The multi-agent dynamical systems theory needed to heal internal conflicts such as auto-immune disorders may not be so different from those needed to heal external conflicts as well, including breakdowns in social and political systems.
We don’t expect to be able to control such large-scale systems, but we think healthy is the best word to describe our desired relationship with them: As a contributing member of a well-functioning whole.
Translation: they don’t know the first thing about healthcare but want the big US healthcare grifting dollars anyway.
Surely they can tell us what they actually worked on in the gaming space in the past few years. Right? Right?
Edit: (I recall hearing stories about roguelike developers using AI to help produce roguelikes, the conclusion was (this was more than a year back btw) it works great if you just need some quick content, dialogue trees, stuff like that. But there also was one person who was trying (and failing) to create a roguelike in which all content was LLM created. Didn’t seem to work well. (Which is funny in a way, as this mirrors the experience of PCG (procedural content generation) in roguelikes a ~decade earlier. Good tool for small parts, doesn’t scale to properly interesting content. I should ask how the LLM stuff is doing now).
Here’s what they write:
AI alignment via the power of videogames:
We’re starting with a singular focus on video game development, because we think that will offer the best feedback loop for testing new AI models. Over the next decade or so, we expect an increasing number of researchers — both inside and outside our company — will transition to developing safety and alignment solutions for AI technology, and through our platform and products, we’re aiming to provide them with a rich and interesting testbed for increasingly challenging experiments and benchmarks.
Healthcare pivot:
Originally, when Encultured was founded as a gaming-oriented AI research company, our immediate goal was to make research progress on human–AI interaction that would ultimately benefit humanity well beyond the entertainment sector. Since then, we’ve considered healthcare as a likely next step for us after gaming.
Couldn’t find any details beyond that. Perhaps one of them read way too much Friendship is Optimal but they didn’t actually having any gaming chops so never got anywhere.
EDIT: More details here: https://www.lesswrong.com/posts/ALkH4o53ofm862vxc/announcing-encultured-ai-building-a-video-game
More details.
Wow, they actually succeeded at their plans. I’m impressed. “we expect to be much more careful than other companies to ensure that recursively self-improving intelligent agents don’t form within our game and break out onto the internet!” Well done!
@V0ldek @sailor_sega_saturn remember all those awful NFT games? I imagine it’s like that, but stupider
