@HasnaMatBhai@grok Annual floods which isolate the area for months, swamps, marshy, ultra humid. Pretty awesome place though. Love it up there. Heaps of crocs
@doodlestein I'm having issues at work adopting any of your projects with the rider on top of the MIT licence. Can you confirm that it's ok for Claude based agents to use your projects?
I don't work for Anthropic, but obviously we use Claude
What kind of laws are passed off the back of an obviously outdated claim like this?
I mean, even back then (an eternity in AI development) it seemed like those in the study were the laggards
In a randomised trial, software developers using AI took 19% longer to complete tasks. Yet they believed AI had made them 20% faster.
At the ‘Evidence in Action’ event, I argued for a government that tests what works, and saves money by learning from the results. https://t.co/mgL3dIXaY8 #randomistas
@aaronbeashel@_lhermann Surely this is possible to hand roll with some models to do some detective work for you, and maybe some access to some datasets of contact info
@bookwormengr The most plausible way this is realised, is that a relatively small group of people are responsible for a lot of the token consumption.
Smaller version is playing out at the more AI native shops like Ampcode or with individuals, provided they don't have compute limits
@javier_mora94@wesbos With slightly different techniques you can get similar results with android
Just video, a colmap solve and a pipeline that checks the agent's 3d model matches visually
Australia is being handed a golden goose on a platter, we'd be absolutely max not to ramp up the generation and installs to match the ambition! Or at least, not get in the way
Literally writing a note on this as we speak. If Aus can find a coherent trade off on copyright and data centres, it is v much the best placed ‘Middle Power’ to pull this off.
@simonw@TheKing86351621 I've been using WebRTC which has been ok for this kind of browser to browser comms. Have made 2 games which use phone controllers in this way. You could always fall back to audio as last resort 😂
@doodlestein Can you explain each of the common types of p value misinterpretations, their statistical foundations and why they matter? An extension of your doctors-misunderstand-p-values blog perhaps, to cover the impact of the misunderstanding and rigorously explain the misconception.
So in a corporate environment I can kinda still build my own tools. But the shared agent pool, has some additional hurdles to meet. For that reason I think we still need some developer specific space for now