Venkatesh Rao 🔹

@vgr.bsky.social

11201 Followers 362 Following

Delver at contraptions.venkateshrao.com

This is the first year in nearly a decade where the new year arrives without the feeling all the planning and resolutioning either being in the shadow of ascendant Trumpism (even in the Biden years) or an effort to studiously and strenuously ignore it. AI has easily marginalized the Trumpist turn of the planet in the attention stakes, despite it being stronger than ever. It now merely feels like an annoying transient problem that doesn’t even really need to be solved, just outlived. It’s not an enduring existential condition to be adapted to. AI not only is that, it has so far been largely a positive turn of history as far as I’m concerned. The right things are thriving or dying because of it. It is pissing off and energizing the right people. It is creating its own new problems but they are ones that seem interesting and worthwhile to solve. I’m happy and eager to adapt to an AI world, where I merely stoically accept a climate-change world and have been implacably resistant to adapting to a Trumpist world. It’s nice to feel the worldwinds of historical change blowing roughly in favorable directions relative to where I want to go for a change. Happy new year! Feels like the cheeriest one in a long time. May the slop be with you!
Write a short story exploring the premise of an alternate set of non-Asimovian 3 laws of robotics, featuring one or more contemporary real robots (like Boston Dynamics, Figure8, Unitree), and based on reasonably realistic near-term assumptions about where the AIs and control architectures of real robots can go (ie, don’t make up magical “positronic brains” — base it on how robotics ML and control actually work). Your pitch should include the 3 alt laws you want to explore, and the premise that will stress them (if you’re familiar with Asimov’s original stories, they usually involve a seemingly impossible violation of the laws that is then explained). Bonus for making one or more laws weird or strange (Asimov’s og ones are pretty logical), but still elegant and effective. For example you could take inspiration from no-fault insurance rules and make it a robot law that some human is always marked as clearly at fault for anything the robot does that goes wrong. Another suggestion — explore laws that operate at different levels of abstraction. Asimov’s originals are all high level and “explainable.” What happens if you work with low-level laws like “every decision the robot takes must be a majority decision of the simpler micro agents constituting it.” Post your pitch as a reply containing a Google doc link open to public comment. You can iterate and refine and steal from each other until the deadline. If you’re new to our bounties process, check out the README thread and ask any general process questions there. https://discord.com/channels/1082444651946049567/1426613444735402116