Writing emails is likely no one's favorite task. With the emergence of generative AI, it's unsurprising that this use case has become widely popular in work settings. Whether we fully use it or just for editing, it remains a growing application of generative AI.

In this trial, we aim to determine which tools excel at this task and establish rigorous assessment criteria covering both technical capabilities and practical applications.

Assessment Criteria

First, we'll test email quality. There's an obvious need to communicate grammatically correctly, clearly, and coherently. The language should use a professional tone, and the structure should be easy to read.

Second, we'll assess accuracy and information integrity. This is paramount to ensure that the AI adheres to the provided information without embellishment. It must avoid fabricated facts or hallucinations.

Third, we'll take a look at relevance and customization. AI must be flexible enough to understand and address specific instructions as well as adapt to various situations.

Fourth, we'll test consistency to maintain a uniform voice across multiple emails. It should also deliver consistent quality regardless of complexity.

Fifth, we'll take into account user experience, which includes user interface, ease of use, and generation speed.

Finally, we'll assess authenticity. We believe this is crucial to match the desired voice and style without having an artificial tone. We'll make sure the candidates write context-appropriate emails that adhere to the user's style guide or past examples.

Nominations and Face-Offs

Completed

Schedule