@AmandaAskell ... I feel like Asimov may have been running a more comprehesive eval suite than modern red teams? Reframing the conversation here... thoughts? Creating Alignment: The Search for Possible Futures https://t.co/kokajHpage
@gdb@claudeai@DarioAmodei They’re embedding it into the DNA of their (public benefit) company, just like privacy was embedded into Apple’s (for better, and worse — ehem, Siri model limitations). Nothing is permanent but a commitment is a start!
Every means-tested program creates two costs:
1.Administration
2.Fraud prevention
Universal Basic Income eliminates both by design. If everyone qualifies, there’s nothing to game.
If I were to roughly define the large, middle band of American politics right now in 10 points, I think it'd look something like this:
1. Relentless focus on solving affordability; "Abundance" x "Cut regulation" agenda
2. Moderate social politics that defend weak/disenfranchised and despise cruelty, but refuse to have ultra-progressive cultural/social norms enforced on them
3. Support work of police/military but also want unapologetic accountability for bad actors
4. Policy of soft American power (supports USAID, being beacons of hope across the globe) but as a disposition is isolationist (no more wars, not dragging us into conflicts in Ukraine OR Venezuela)
5. Intense border security and zero tolerance for people here illegally committing crimes, but also makes it easier to come here legally to work/live/immigrate and welcomes newcomers who want to be American & embrace our society/culture as it is
6. Wants govt funding + private sector coordination toward diseases of despair (addiction, suicide, social media addiction)
7. Cares about birth rate and promotes pro-family policies to make having families easier; calls on not just govt but also faith orgs in pursuit of promoting more + bigger families
8. Environmentalism that cares about climate change but also embraces nuclear + technological solves, not focused on limiting/banning energy sources - wants to win on the merits in the capitalist system
9. Believes in expanding ACA+Medicaid at state and federal level to address health insurance funding, but also emphasizes pressures on hospitals/PBMs who drive up cost
10. Non-negotiable reforms to Social Security, Medicare, and military+govt waste to address our insane debt which will one day bring the economy to ruin
This wouldn't be MY 10-point plan exactly, but I'd definitely prefer this rough outline over either major political tribe we have now. It's incredible how under-represented this band of voter is, yet how common they are in my personal and professional life (and among my many thousands of readers).
Where is the politician or party ready to draw these lines? I think in a post-Trump era they'd be incredibly successful.
@DGisSERIOUS Agreed. I think we (SCOTUS? Congress?) really need to think deeply about how/whether we delineate the right to free speech from the right to free & limitless reach. The magnitude and impact of this distinction in the 21st century was unimaginable in the 18th century.
Remember the #NoKings rallies? Meanwhile Kash Patel is writing children's books idolizing Trump as a King and himself as the hero who saves the day. You can't make this shit up.
@Jackbmeyer@hillaryvipond Kirk was both a product and perpetuator of a win/lose political culture. This tragic outcome is a reminder that we must do better if we are to overcome the challenges of this era.
@Jackbmeyer@hillaryvipond We need intellectual humility, a willingness to be wrong and refine our perspectives — not to start from a place of “prove me wrong” but from a place of “help me understand” and “let’s grow together”.
@Jackbmeyer@hillaryvipond Our society and system rewards this kind of performance. I think there’s an element of self-deception that debate amidst disagreement and engagement with opposing viewpoints is sufficient. It is necessary, but not sufficient. We need more than “engagement”.
People leaving regular companies: Time for a change! Excited for my next chapter!
People leaving AI companies: I have gazed into the endless night and there are shapes out there. We must be kind to one another. I am moving on to study philosophy.
Paper from OpenAI says hallucinations are less a problem with LLMs themselves & more an issue with training on tests that only reward right answers. That encourages guessing rather than saying “I don’t know”
If this is true, there is a straightforward path for more reliable AI.