Just curious... What models are you talking about?
Models as in computer models. These are typically mathematical models. They get different labels based on the latest jargon and/or underlying technique used.
These are some common models:
- Physical models (simulate individual pieces of the problem)
- Statistical models (use distributions to simulate the system individually or systematically)
- Vision models (specialize in image and video analysis to identify objects, events, etc.).
- LLM models (language based models based on a variety of techniques and interacted with via text)
- ML models aka Machine Learning Models (usually have a reward function to direct training and execution of the model to incentive a particular behavior)
- RL models aka Reinforcement Learning Models (specialization of ML models that not only have a reward function but also value functions and a few other details that allow "learning" independently from pre-defined answers)
- AI models aka Artificial Intelligence models (a catch-all term and rebranding of LLM, ML, and RL models)
- Generative AI models (specialize in creating images, videos, and/or text)
In this case, Lucid would be developing visual models at a minimum. They could also be developing physical or statistical models to enhance the outputs of the visual models. Finally, they should be layering a final model to take actions on the information produced from those models. That could be RL, ML, or something else entirely such as a system of pre-defined rules and actions.
