Can generative AI understand a continuous, messy world?
One participant argues that generative architectures, especially those producing discrete symbols, are ill-suited to understanding reality. A reply offers floating-point numbers and transformer-based video generators as counterarguments.
TLDR
The disagreement centers on whether generative architectures can handle real-world complexity. One participant argues they are the wrong approach—particularly when generating discrete symbols—because the world is high-dimensional, continuous, noisy, messy and largely unpredictable. A reply counters that enough discrete symbols can approximate continuous values, citing floating-point numbers. It also argues that transformer-based video generators seem to capture much of the messiness of real-world inputs.
Can generative AI understand a continuous, messy world?
One participant argues that generative architectures, especially those producing discrete symbols, are ill-suited to understanding reality. A reply offers floating-point numbers and transformer-based video generators as counterarguments.
TLDR
The disagreement centers on whether generative architectures can handle real-world complexity. One participant argues they are the wrong approach—particularly when generating discrete symbols—because the world is high-dimensional, continuous, noisy, messy and largely unpredictable. A reply counters that enough discrete symbols can approximate continuous values, citing floating-point numbers. It also argues that transformer-based video generators seem to capture much of the messiness of real-world inputs.