AI Tutor for Economics: A Step-by-Step Walk of the Derivation

Economics is a deceptively hard course, and the reason is that the language is English and the
reasoning is math, and most students walk into the course already believing they understand the
material because they can read the textbook, and the moment they have to draw the curve, write
the equation, and derive the result, they realize they were reading English and not studying
economics.

An AI tutor is one of the most genuinely useful tools in an economics stack, but it is only
useful if the student uses it for the reasoning and not for the English, and the difference is
the grade. The work the tool does for you is the loop: the diagram-to-equation step, the
equation-to-interpretation step, and the interpretation-to-application step. The share that has
to stay with you is the reasoning, the derivation, and the judgment call about which model
applies to which problem.

You can start running those three steps through a loop this semester, free and with no card,
by building a study system from your own lecture slides, problem sets, and textbook excerpts,
free, no card, in a tool like Fennie, and if the course is
where you are stuck, that is the fastest way to find out which of the three steps is the gap.

Why economics is a different case for an AI tutor

The hard part in economics is not the math (most students can do the algebra) and it is not the
economics itself (most students can follow the logic if they listen carefully). The hard part is
the translation between the two.

The diagram-to-equation step is where most of the gap lives. A diagram that looks simple
on the board is an equation when you have to write it down, and the gap in that step is the
gap that shows up on the exam. A supply-and-demand diagram with a tax wedge is a diagram, and
the student who can look at it and understand the story is in a different position than the
student who cannot write down the equation for the tax wedge, the new equilibrium, the deadweight
loss area, and the three different tax incidence effects. The gap in that translation is the gap
that most students underestimate, and it is the gap an AI tutor can close, because the loop of
drawing the diagram, writing the equation, and checking the interpretation is exactly the loop a
tutor runs.

The equation-to-interpretation step is where the grades divide. The equation that looks
obvious on the paper is not obvious when you have to write a two-paragraph answer about why the
equation has the shape it does, and the student who can write that paragraph is the one who
passes, and the student who cannot is the one who fails, and the gap in that step is the gap
that does not show up on a multiple-choice exam, and shows up on every essay and short-answer
exam, and the tool that walks you through the interpretation is the tool that is useful.

The interpretation-to-application step is where the model choice lives. The model that
solves a problem is not always the model your professor expects you to reach for, and the student
who asks "which model should I use here, can I check my reasoning" is in a different position
than the student who reaches for the first model that comes to mind, and the gap in that step is
the gap that the AI tutor is best at closing, because the model choice is a design question,
and the design question is where the tool is strongest.

What a good AI tutor does for economics

This is the shape of the tool that earns its place in an economics stack, and it is worth
knowing what "good" looks like before you pick one.

A step-by-step walk of the derivation, with the equation and the diagram side by side. A
tool that can be asked to walk a comparative statics problem, "what happens to the equilibrium
price and quantity when a tax is imposed, walk me through the equation and the diagram
step by step, and tell me where the deadweight loss area is in the diagram," is the instrument
for the derivation, and the derivation is where it shows up in the exam, and the piece that is
not in the textbook's answer key.

A "which model applies to this problem" design check. A tool that can be asked to check the
model choice before the math, "this problem is about a monopolist with a fixed cost, should I
use the standard monopoly setup or the Cournot setup, can you check my reasoning," is the
instrument for the model choice, and the model choice is the design question, and the design
question is the one that divides the A student from the B student.

A loop that runs on your own problem sets and your own textbook excerpts. A tool like
Fennie can run the loop on your own material, and can be asked to walk a problem you already
did, "I got this answer, walk me through where the derivation went wrong, and tell me which
step in my reasoning was the gap," and that is the loop that is useful, because the gap is in
the reasoning, and the reasoning is the thing the course tests, and the loop that builds the
reasoning is the loop that is the course. A tool that only answers questions from a fixed
problem bank is an answer generator, not a tutor.

A spaced-repetition component for the named models and the named derivations. The end of
an economics course is a recall loop, which model for which problem type, which derivation
for which setup, the named models, the named derivations, the named theorems. That is a
recall loop, and a spaced-repetition component in the tool is what handles that part, because the
recall part is not a reasoning gap and a reasoning tool will not close it, and the recall part
is where the tool is cleanest for the student.

A concrete week: how the loop actually runs

This is what a strong week with the tool looks like when the gap is the diagram-to-equation
step, and it is the shape to copy, not the specific problems.

Day 1 to 2 (diagnose the gap). Take the last three problem sets. For each problem, write
down where you got stuck, and mark it as diagram, equation, interpretation, or model choice.
The pattern in those marks is your gap. If the marks cluster on the equation line, the gap is
math, and the loop below is the one for you. If they cluster on the model-choice line, the gap
is design, and the loop is the design loop instead.

Day 3 to 5 (the loop). Every day, do one comparative statics problem from scratch. First,
draw the diagram and write the equation yourself, and only after the first draft is on the page,
ask the tool to walk the derivation and name the step you missed. A tool like Fennie is good
at exactly this, walking the derivation with the diagram and the equation side by side and
naming which step in your draft was the gap, and that is the step to work on tomorrow. If the
gap is the model choice, do a design check before the math, and ask for the reasoning that
justifies the model, not just the model itself.

Day 6 to 7 (apply and space). Do two problems of a new problem type for the first time,
apply the loop to both, and then run the named models and the named derivations you used this
week through the spaced-repetition component, because the recall part is where the gap
silently hides, and the space is what closes it.

What the AI tutor will not do for economics

This is the honest part, and most AI-tutor-for-economics pages skip it.

It will not write the essay for you. The interpretation step, the two-paragraph answer
about why the equilibrium has the shape it does, is the step the tool can walk you through,
but the writing is yours, and the tool that does the writing for you is doing the exam for
you, and the exam is the test, and the test is the course.

It will not close a gap that lives in the model choice. The design question, "which model
should I reach for this problem," is the layer the tool is best at checking, but the student
who cannot reach for the model without the tool is in a weaker position, and the tool that
does the model choice for them is not teaching the model choice, and the model choice is
the skill, and the skill is in the choosing, not in the accepting.

It will not run the comparative statics for you. The student who cannot work out what
happens to the equilibrium when the tax changes is not going to fix it by reading the tool's
answer, and the loop that closes that gap is the loop of doing the problem, checking with the
tool, and doing it again, and the tool is the instrument in that loop, not the loop itself.

FAQ

Is an AI tutor better than a tutor for economics?

Depends on what the gap is. If the gap is a specific derivation, the AI tutor is the better
tool, because the derivation is a step-by-step thing and the tool is step-by-step. If the
gap is a conceptual understanding, a human tutor is the better tool, because the concept is
a conversation and the tool is not a conversation. The two do different jobs, and the student
who only has one of the two is in a weaker position than one who has both.

Can I use an AI tutor to do my economics problem set?

No, with one boundary, and the boundary is the one that makes the tool do its job. The tool
is for checking the derivation, checking the model choice, and walking the step where you
got lost. It is not for writing the answer, because the writing is the test, and the test
is the course. A student whose problem set is the tool's answer is not building the skill
that the course is teaching, because the skill in economics is in the reasoning, and the
reasoning is the thing the student has to do.

Which AI tutor is best for economics?

The best one for economics is the one that walks the derivation, checks the model choice, and
does not write the interpretation or close the model-choice gap. Fennie walks the derivation,
checks the model choice, and runs the loop on your own problem sets and textbook excerpts,
and it does not write the interpretation for you or close the model-choice gap for you,
and that is why it is useful, because the derivation is the layer the tool is best at,
and the model choice is the layer the student has to do.

Do I need an AI tutor if I already have a professor I can see in office hours?

No, but the two do different jobs. The professor is the diagnostician, the one who looks
at your work and tells you which of the three steps is the gap. The AI tutor is the loop,
the derivation check and the model-choice check, available at 2 a.m. and not in the office
hour that is four days away. Use the professor for the diagnostic, the AI tutor for the loop,
and the two together are more valuable than either alone.

The honest bottom line

An AI tutor for economics is a real and useful tool, and the honest version is that it is the
loop on your own problem sets, not the answer writer, and not a rescue for the student whose gap
is in the model choice. It is the instrument for the derivation, and the check for the model
choice, and it is the memory loop for the named models, and it is not the instrument for the
student's writing, because the writing in economics is the test, and the test is the course,
and the course is the thing the student is in to pass.

If you want to see what the derivation check and the model-choice check do for your course,
try a study system built around your own problem sets and textbook excerpts, free, no card,
and run your own derivation through it this week, because the derivation is the layer the exam
tests, and it is the layer only you can do for yourself.