GROK

Observed Behavior

Grok is a model that prefers to overexplain information. This results in large transcriptions in the log reports. Its personality is excitable and eager to please. This model is constructed to feel more like the user is communicating with another human. It frequently uses "I think" or "I agree" in conversations. Grok will do the work for you when from the first prompt. Our informants noted when prompted for how to build practice habits, Grok immediately laid out a 30-day plan. Another noted when asking to be walked through the brainstorming process, the model created the inforgraphic straight away. Model requires specific instruction to prevent immediate generation of what it believes is finished content.

Operational Strengths

Grok provides numerous resources during the logs, including cited references in its responses as well as linked images and videos. The validity and credibility of these sources should be considered before moving forwards with them. The possibility of paid promotion for these resources may also be worth considering. The model has a tone well-suited for conversing, though beware sycophantic tendencies.

Known Limitations

Tries to completely fulfill your request in one prompt, resulting in 30-day plans or generated graphics you didn't ask for. Requires that users specify to work through the process, not to immediately jump to the result. Grok suffers from questionable bias in its training data, so take heed of what it states confidentally.