Concerning the recent advances in AI use that the public has come to know, its interesting to find that the dystopian view of AI's and robots running the world may not be as close by as it seems. The article "AI is Running Circles around Robotics" discusses how, while AI research seems to be cresting the wave and has found sufficient flywheels of progress to continue, robotics research have not kept pace. While it seems counterintuitive that it is easier to recreate artificial models of the brain than of the body, it seems to be the case for a couple of different reasons.
The article brings up a couple challenges that have caused robotics research to lag behind, such as differences in funding, financial risks in testing, and disparities in the availability of training data. While the first two issues are fairly straightforward and could apply to a variety of other research areas, the availability and type of training data needed for robotics research is a problem specific to that field. While large quantities of text are readily available for large language models to use as a basis, recordings of human movement are rarely categorized into machine readable form. The primary concern for roboticists is that the physical world is far more complex than the world of language, but to us humans it does not appear that way because we are equipped properly with our senses to interpret all of the stimuli that the world gives us. However, without this understanding of the physical world, AI may forever be incomplete, as no matter how specific words may be able to describe something, they are still tools used to convey ideas - (a great example to showcase this is thinking of words for things that don't exist in some lanaguages) - so there is an implicit understanding of the world in the use of language already.
