Ingenuity flies...

...and makes the first powered, controlled flight by an aircraft on another world.
The flight was the first of five that are planned and this one was mostly proof of concept. The vehicle spun up its rotors, took off, rose to a height of 3 meters and hovered there for a few seconds before landing.
An historic moment in space exploration. Hopefully, future flights of Nasa's martian helicopter will be just as successful.




Comments

Post a Comment

Popular posts from this blog

10 minutes of...

Image
...breathable air; about 5 grams of oxygen. It doesn't sound much and the limitless supply of breathable air we have on earth means it doesn't sound an impressive quantity. Yet the location and production of these 5 grams of oxygen make it important. In between the first two flights of the martian helicopter, Nasa announced the Perseverance rover's MOXIE instrument made a less visually dramatic but potentially more useful contribution to manned space exploration; producing breathable air from the thin, carbon dioxide-heavy, martian atmosphere. The nickel alloy MOXIE instrument achieved this by taking in samples of Martian air, heating it to 800 degrees celsius and thermochemically splitting the CO2 into oxygen and carbon monoxide. The latter gas was vented and the oxygen stored. Moxie can produce twenty minutes of breathable air per hour, per person. So three modules could sustain an astronaut indefinitely. Assuming continuous supply of power and reliability. There will be
Read more

Google brings seed AI...

Image
...one step closer, according to an article in  nature.    The development they are reporting involves the use of reinforcement learning to dramatically increase the speed at which silicon chip floorplans can be designed, especially the floorplanss of Google's own Tensor Processing Units (TPU's). Since TPU's are themselves optimized for machine learning algorithms, this work represents a potentially important development in the field of artificial intelligence, since it could herald the beginning of what Roman Yampolskiy termed,  seed AI . This is the scenario where a human-created AI possesses engineering capabilities that surpass those of the very humans that created it, and would thus be able to continually improve its own design, leading to ever more advanced hardware.   A  CNBC  article covers the same story, while Yann LeCun, Facebook's chief AI scientist, described it as 'very nice work.'      
Read more

100 trillion parameter...

Image
...gpt-3 models will be a thing by 2023, according to Nvidia CEO Jensen Huang. He made the prediction in his GTC keynote speech. I've posted the youtube video and it should start at the relevant timestamp. As Jensen points out, the human brain has 125 trillion synapses, so by some measures, especially in the field of language processing, machines could soon be getting close to human-level capability. Document summaries, phrase completion and the ability to produce code from plain English, are a few examples. As Jensen makes clear elsewhere in the video, computers writing the software to run computers is a major goal of the research. With the ability to both understand language input and, in response, code their own software, computers will be showing many of the hallmarks of intelligence while outperforming people numerous tasks. Assuming they're not too busy mining bitcoin.
Read more