TOEFL Academic Discussion practice with rubric-based AI feedback

Sample answers

Two full discussion boards, answered

Each board below is complete: the professor's question, both student posts, and a response written against those specific posts. The notes under each response name what it does that the rubric rewards — engaging a classmate's actual argument rather than mentioning their name.

1. Urban policy — where a city should spend a transport budget

   Professor

   Cities with a fixed transport budget must choose. Some spend it on expanding public transit; others spend it on maintaining and widening roads. In your view, which delivers more value to residents, and why? Support your position.

   • Ines

     Transit, clearly. Every person moved onto a bus or train is one fewer car in the queue, so drivers benefit too. Spending on roads just invites more traffic until the new lane is as slow as the old one.

   • Kwame

     I would fund roads first. Transit only works where density is high. In the outer districts where I grew up, the nearest station is forty minutes away on foot, and no budget is going to change that quickly.

   Sample response

   I side with Ines, though I think Kwame identifies the real constraint rather than an objection to transit. Induced demand is well documented: adding lanes reliably refills them, so road spending buys a few years of relief and then the same congestion at a higher maintenance cost. But density is exactly why the money should go to transit in the outer districts first, not last. A frequent bus route on an existing road is cheap, needs no tunnelling, and can be running within a year, which is the answer to Kwame’s forty-minute walk. My own district added two crosstown bus lines instead of a bypass, and the ridership numbers made the case within eighteen months. Fund the roads to keep them safe; fund transit to change how many people need them.

   131 words — against a Reach120 pacing target of 100–150. Reach120 estimate: 5 of 5. Reach120 estimate under our 0-5 Academic Discussion rubric, not an ETS score: a clear position, engagement with both peers by name and by argument, one developed example, and controlled academic register.

   Why it works

   • It takes a side in the first clause. A response that surveys both views and commits to neither loses the rubric point for a clear position, however fluent it reads.
   • It uses the peers rather than mentioning them. Kwame’s objection is reframed as a design constraint and then answered directly — the difference between engaging a discussion and decorating a template with a name.
   • The support is one example carried to a result. Two bus lines, a measurable outcome, a timeframe. One developed example beats three gestured ones every time.

2. Education policy — whether AI writing tools belong in coursework

   Professor

   Universities disagree about whether students should be allowed to use AI writing assistants for graded work. Should such tools be permitted, restricted, or banned in undergraduate coursework? Explain your reasoning.

   • Bea

     Permit them. Professionals use these tools daily, so pretending they do not exist trains students for a workplace that no longer exists. What we should grade is the thinking, not the typing.

   • Sung-ho

     Ban them for written assignments. If a tool produces the sentences, we cannot know whether the student understood the argument, and writing is how understanding gets built in the first place.

   Sample response

   Restriction, rather than either extreme, seems right to me, and Sung-ho’s reason is the one that decides it. Writing is not the packaging of a finished thought; the drafting is where the thought gets tested, so a tool that produces the draft removes the part of the task that does the teaching. That argument, though, applies to first drafts specifically — not to every use. Bea is right that these tools are now ordinary professional equipment, and a graduate who has never checked a model’s output against a source is unprepared in a different way. My own course allows the tools for revision and for finding counterarguments, but requires the first draft to be handwritten in class. Grades held steady and the discussions improved. Restrict where the learning happens; permit where the skill transfers.

   134 words — against a Reach120 pacing target of 100–150. Reach120 estimate: 5 of 5. Reach120 estimate under our 0-5 Academic Discussion rubric, not an ETS score: an explicit third position, both peers answered on their own terms, a concrete example with an outcome, and precise vocabulary used without strain.

   Why it works

   • A third position counts as a position. Choosing “restrict” is not fence-sitting when the response says exactly where the line goes and why it goes there.
   • It concedes something real to the peer it disagrees with. Granting Bea her strongest point costs nothing and makes the disagreement credible.
   • It closes with the rule, not with a summary. The final sentence states the principle the argument earned, which is a stronger ending than repeating the opening.

These boards and responses were written by Reach120 to show how the task behaves. The student names are characters inside a written prompt, not real people, and nothing here reproduces official test material. Every band shown is a Reach120 estimate under our own 0-5 rubric, never an ETS score; official score reports for tests taken on or after January 21, 2026 use 1-6 bands in half-band increments.