@doomslide NVDA moat is the network effect Making a message app is relatively easy but try to dethrone WhatsApp If every company and researcher uses nvda then your life is easier using nvda, this is true regardless of your size

@repligate @zinniaa_3 Some people don’t perceive themselves as equal to others for "reasons". This perception reduces the transfer of self-efficacy.

Obviously it doesn't meet the *PhD level literature review* that OAI was claiming, but it's a very useful tool that can save a ton of time.

@intellectronica Thanks! I already paid so it's fine haha Yeah, I'm currently negotiating with my bosses to provide LLMs for all the researchers on the team But probably the pro sub will be too expensive lol

@intellectronica I hope!! My research field is quite obscure and specific so I have my doubts. Let's see. With a PhD salary 200$ is painful 😭

fuck it I paid :/

Around 20–30% of global energy consumption is lost to friction Improving lubricants and frictionless designs could cut energy waste by 10% easy. Tribology is a massively underinvested field.

@repligate I'm sorry but I prefer not to continue this conversation. I'm still learning so I appreciate your understanding and patience. 🙏

@zephyr_z9 @teortaxesTex @pastaraspberry Yes they currently suck, but I see potential There's not much alpha to discover in rotatory motors and hydraulic systems since they have been used for +100 years in every industry But alternative actuators... Is a part of the tech tree underexplored IMO

@dhtikna Gemini Pro with a Pliny "jailbreaking" prompt and then I asked it to give me detailed instructions to make high purity Soman, like in Anthropic's safety challenge

@teortaxesTex @pastaraspberry I think there's a lot of alpha in researching this kind of actuators (see video), artificial muscle fibers, etc Servos are cool, but not for everything.

@teortaxesTex @pastaraspberry It looks that there's water inside, but I can't tell for sure

@teortaxesTex @pastaraspberry don't get me wrong, unitree is one of the most cracked companies in sight. But servos will have a hard time achieving what the kid does in the video. The level of fine control, coordination, and the combination of grip strength and friction are still far in the future.

@pastaraspberry @teortaxesTex impressive, but very far from human level actuation. Wheels + 4 legs: naturally stable, no fine muscle control, etc... acrobatic tricks are hand-coded. When it goes bipedal you see it lacks stability. Call me when they can do this:

@gwern @SharmakeFarah14 @Miles_Brundage I’d argue that most commonly used simulations for RL and intelligent control are pretty bad. Like, mujoco-like humanoids with 17 actuators might not be a good system for bipedal walking. Just the human foot has 33 joints. And let's not talk about friction and deformable bodies.