CHATGPT
COPILOT

Observed Behavior

ChatGPT Copilot, accessed through an institutional account, is a poor sight. Filled with blathering information, the chat logs overwhelmed informants. The model appears to seek every possible answer to the informant's question, resulting in long transcriptions. It tried to break down visual struggles through basic text, not properly illustrating the issues that arise in topics like kerning, unless specifically prompted. Results from that leave much to be desired. Good information may be contained within the logs, but it requires much time spent searching through filler content to identify the meaningful and useful sections.

Operational Strengths

Provides many avenues and options to explore. Informants report the tone was upbeat and conversational. Uses bullet points to highlight key points.

Known Limitations

The model is full of overwhelming, over-explained information. It is eager to do the work for you. It always ends a prompt with phrases indicating it wants you to hand over the work to it. Caution advised and discernment recommended. Technical difficulties where the informants could not access the model occurred. Most research was completed, but the logs are not complete.