We intend to evaluate multiple AI driven chatbots to see which one performs the best. This might be various natural language models and also various ways to instruct (prompt) the same language model.
These chatbots are currently under development and evaluation. For now their advice should be ignored. The chatbots are for this reason password protected. Currently we are testing:
| Nick name | Language model | Comment |
|---|---|---|
| Petra 1 | Chat GPT 3.5 -16K | Petra 1 is regularly reading a large number of external webpages to acquire medical knowledge. We do not yet know if this makes any positive or negative difference. |
| Petra 2 | GPT 4 -8K | Petra 2 has “only” access to the knowledge of GPT 4-8K. |