The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Are LLMs dead or alive?
A human-centric perspective
Crystalline and a beam of light
Imagine a crystal that refracts light in intricate, dazzling patterns to create a visual spectacle.
An LLM can be likened to such a crystal, with the prompt serving as the beam of light.
Fixing the prompt, the random seed, and the runtime parameters, is analogous to holding the beam of light perfectly still: it will yield the exact same projection every single time, without deviation, for the rest of eternity.
A system where a set of inputs yields a fixed outputs is a static one that is more artifact than organism. In this sense, an LLM is far more dead than alive.
By changing the angle and distance of the light source, varying the model’s prompts, we can create the illusion of dynamic outputs. But this is more a trick of the light than living thought.
Aliveness beyond human perception
Limited by our human-centric views, we tend to recognize life only in things that closely resemble ourselves.
It is also possible to interpret some entity as being alive by their ability to generate new information.
By this measure, even a rock may be considered alive. As are stars in their slow galactic dance.
This does not rule out alive AGIs in the future. Because human consciousness itself arises from physical substrates, living synthetic intelligence remains entirely possible.
- Downloads last month
- 30