Frequently Asked Questions

How does EvalUGrade work? I don't know that I trust a complicated attempt at Artificial Intelligence (AI) such as ChatGPT to evaluate work.

Neither would we! Our patent pending method of grading is mostly heuristic, relying on the minimum necessary amount of AI interwoven into a much larger engineered solution.

What sort of work are you capable of evaluating?

We can evaluate Math up to Calculus II, including algebra and trigonometry. Additionally we can do baby physics. We'll be expanding further as quickly as we can.

Why would anyone use your specialized service over something general like ChatGPT?

Despite it's promises Generalized PreTraining (GPT) does not confer infinite benefits and functions well to a pretty specific set of problems. As far as the domain containing the only AI part of our application, GPT is not performant. As an example of ChatGPT in the non-AI component of our application, it may properly perform two digit multiplication but as the digits which one deals with expand it is no longer correct. One must remember that it finds answers in a manner akin to how a librarian might find a book, however the book is a fuzzy remembrance etched in ink that depends on a set of interdependent inkblots.

How Accurate is EDUP.AI?

We will tell you if your solution confuses us. We require clarification if we cannot deduce how you produced a step from the previous one.

There is a symbolic engine that backs EDUP.AI and while we occasionally have bugs they are not so suitable as to only effect our correctness. You'll know when we're broken.