TARA Writing Task Feedback: Two Essays Compared

A TARA essay can answer every part of the question, read clearly and still need substantial improvement.

That is exactly what happened across two real, anonymised practice examples examined for this guide. One response addressed all three parts and had a logical structure but received an indicative practice score of 46/100. The other developed the same underlying ideas more fully and received 92/100.

The difference was not simply vocabulary or essay length. The detailed feedback identified a deeper distinction: one response presented relevant ideas, while the other developed those ideas into a reasoned argument.

This guide uses both attempts to show what the TARA Writing Task requires, how ExamPrepIQ's independent AI feedback works and how to turn TARA writing feedback into a focused revision plan.

Independent practice note: The TARA Writing Task is not scored by UAT-UK. ExamPrepIQ's scores and feedback are independent, indicative practice tools. They are not official TARA results, do not reproduce a university's assessment process and cannot guarantee any admissions outcome.

What Is the TARA Writing Task?

The TARA consists of Critical Thinking, Problem Solving and a Writing Task. According to the current official UAT-UK TARA overview, the Writing Task lasts 40 minutes. Candidates choose one essay-style question from three and must not exceed 750 words.

The official TARA Question Guide for October 2026 and January 2027 explains that questions typically present a short statement and ask candidates to:

1. explain what the statement means;
2. give a reasoned argument against it; and
3. discuss the extent to which they agree.

UAT-UK does not award a score for the Writing Task. Instead, a copy of the candidate's response is sent to the relevant universities. This makes it especially important to practise the actual thinking behind the essay, not merely chase an unofficial number.

For the complete section rules, see our TARA syllabus and format guide.

The Practice Statement Used in This Comparison

Both ExamPrepIQ attempts responded to the same statement:

“The increasing reliance on artificial intelligence in education will ultimately diminish critical thinking skills among students.”

Both responses were recorded as fully addressing the three requested parts: interpretation, counter-argument and evaluation. Yet their indicative results differed considerably.

Two TARA Writing Attempts Compared

For teaching clarity, this article presents the lower-scoring response first and the stronger response second. This creates a useful baseline-to-stronger-answer progression without claiming that this was the order in which the submissions appeared in the app.

The comparison shows what changes between the two levels of performance: both responses had clear organisation and relevant ideas, but Response B added definition, causal reasoning, developed examples and explicit weighing.

Read Both Complete TARA Writing Responses

For transparency, both anonymised responses are reproduced in full below. The wording is unchanged; only headings and paragraph breaks have been formatted for easier reading on the web.

The Complete AI Feedback Record

The following record makes the grader's decision explicit. It includes the overall finding, the diagnosis for every dimension, the missed opportunity and the action given to the writer. Short quotations come from the complete responses reproduced above.

Response A: complete feedback record

Overall feedback: 46/100

Task achievement

The grader confirmed that all three parts were addressed. However, it described the 283-word response as severely under-developed. It said the components of an answer were present but were not explored in enough depth to create a convincing argument. The report explicitly stated that Interpretation, Counter-argument and Evaluation were capped at 40 because each part was so brief.

Greatest strength

Clarity and organisation. The response moved logically from explanation to counter-argument to evaluation, and the points were easy to follow. The calculator comparison was identified as direct and effective.

Key improvement

Add depth. The grader said the response read like a plan or summary rather than the completed argument. It asked the writer to explain how AI's functions support or undermine critical thinking instead of only listing what AI can do. For this response, it suggested that approximately 500–600 words would provide room for development.

Checklist question

“Have I developed each of my points with reasoning and examples, or have I just listed them?”

Interpretation & Meaning: 40/100

Diagnosis

The opening was a competent paraphrase, but it was too brief and accepted the prompt's central terms without interrogating them.

Evidence identified by the grader

The explanation that excessive reliance could cause students to lose the ability to think deeply was clear. Defining critical thinking through struggle with difficult concepts was also a useful starting point. The response did not, however, explore what counts as “too much” reliance or what “ultimately diminish” implies.

Missed opportunity

The writer could have questioned whether the word “ultimately” describes an inevitable long-term effect or a temporary risk during adjustment.

Action items

1. Identify the two or three most important abstract terms before writing.
2. Explain what each term means in this context.
3. Consider whether a term has more than one interpretation and state which interpretation the essay will use.

Counter-argument & Reasoning: 40/100

Diagnosis

The response identified several relevant counter-argument points but presented them as assertions rather than developing them through reasoning.

Evidence identified by the grader

Finding articles, summarising chapters and proofreading were relevant examples, but the response did not fully explain how the saved effort would lead to better critical thinking. The personal-tutor point similarly stated that AI could help understanding without connecting that understanding to more sophisticated analysis.

Missed opportunity

The calculator analogy could have supported a complete paragraph explaining how automating lower-order tasks frees cognitive resources for problem-solving and synthesis.

Action items

1. Follow each main point with sentences beginning conceptually with “This is because…” or “This matters because…”.
2. Develop two points in depth instead of listing three brief points.
3. Present the point first, explain it and then give an example that supports the reasoning.

Evaluation & Judgement: 40/100

Diagnosis

The response stated a clear, qualified position, but the conclusion felt asserted rather than earned because the preceding arguments were too brief.

Evidence identified by the grader

“I agree with the statement to a moderate extent” directly answered the question. The final distinction between using AI to avoid thinking and using it to think more deeply also adjudicated between the two sides, but did not explicitly weigh their relative strength.

Missed opportunity

The conclusion could have compared the risk of passive learning with the benefit of AI as a guided personal tutor and explained which consideration was more persuasive under defined conditions.

Action items

1. Begin the evaluation by stating the central tension: “On one hand… but on the other…”.
2. Use explicit weighing language such as “The argument against is more persuasive because…” or “The risk is outweighed by…”.
3. Make the stated extent of agreement follow from that weighing process.

Organisation & Expression: 72/100

Diagnosis

The essay was well organised and clearly written. Its main expression-related limitation was that extreme brevity left too little room to develop its ideas.

Sentence-level feedback

The grader showed how to combine the sentences about intellectual struggle and instant answers into a sharper cause-and-effect statement. It also suggested using an em dash to emphasise the distinction between AI as a starting point for research and AI as a final answer.

Missed opportunity

Each of the three existing paragraphs could have become the topic sentence for a more substantial section. The plan was clear, but it had not been expanded into developed content.

Action items

1. Treat the existing paragraphs as an outline and add three or four development sentences to each main idea.
2. Aim for at least 450 words in the next practice response to create room for development.
3. After establishing the structure, spend the remaining time adding depth and reasoning rather than more undeveloped points.

Response B: complete feedback record

Overall feedback: 92/100

Task achievement

The grader found that all three parts were fully and effectively addressed. It said the response explained the statement, provided a robust counter-argument and reached a nuanced evaluation, with no part of the task missed.

Greatest strength

The report highlighted the essay's sophisticated and balanced structure. It identified “The tool itself does not dictate cognitive decline; the pedagogical design does” as the pivot that resolved the tension between the two sides.

Key improvement

The grader asked for slightly more concrete detail. In particular, the “debate partner” example could have shown what the interaction would look like, such as a student entering a thesis and asking the AI to generate objections from a defined theoretical standpoint.

Checklist question

“Have I identified the single most important condition or factor that decides the extent of my agreement in part (c)?”

Interpretation & Meaning: 90/100

Diagnosis

The explanation moved beyond paraphrase by defining important abstract terms and explaining the causal mechanism implied by the statement.

Evidence identified by the grader

The definition of critical thinking as a skill developed through “productive intellectual struggle” established a framework for the analysis. The explanation that AI could bypass that struggle and outsource cognitive processes showed how and why the negative effect might occur.

Missed opportunity

As an advanced extension, the response could have distinguished between types of AI, such as generative and analytical tools, and asked whether the statement applied differently to each.

Action items

1. Define the most important abstract term in the statement.
2. Explain the causal mechanism that the statement implies.
3. Use the interpretation to establish the central tension the essay will explore.

Counter-argument & Reasoning: 92/100

Diagnosis

The counter-argument contained several distinct points supported by logic and targeted examples.

Evidence identified by the grader

Describing AI as an “advanced cognitive scaffold” gave the section a strong central analogy. Explaining that AI can automate lower-order tasks and release mental bandwidth for higher-order analysis supplied a clear mechanism rather than a bare assertion.

Missed opportunity

The response could have considered new critical-thinking skills created by AI use, including effective prompting and evaluating algorithmic bias.

Action items

1. Organise the counter-argument around a central idea or analogy.
2. Develop at least two distinct lines of reasoning against the statement.
3. Use targeted examples, such as the calculator analogy, to make the reasoning concrete.

Evaluation & Judgement: 90/100

Diagnosis

The conclusion directly answered the extent question and reached a nuanced judgement by weighing the arguments already made.

Evidence identified by the grader

“I agree with the statement only to the extent that…” clearly qualified the position. “The tool itself does not dictate cognitive decline; the pedagogical design does” identified the factor that decided between the competing arguments.

Missed opportunity

The response could have added a brief forward-looking implication—for example, that assessment may need to focus more on students' inquiry process and critical engagement with tools than on the final product alone.

Action items

1. State the level of agreement explicitly.
2. Identify the single most important factor determining which side is stronger.
3. Make the final judgement a direct consequence of the argument and counter-argument developed above.

Organisation & Expression: 95/100

Diagnosis

The grader described the essay as disciplined, clearly structured, concise and precise.

Sentence-level feedback

It suggested combining two sentences about outsourced cognitive processes and dependency to make the cause-and-effect relationship more direct. It also tightened the transition from “The tool itself does not dictate cognitive decline” to the explanation of pedagogical design.

Missed opportunity

The report found little to criticise. It noted that headings were helpful but optional, and that a response under 500 words can still be highly effective when every sentence performs an argumentative function.

Action items

1. Continue planning the structure before writing.
2. Use precise analytical vocabulary only where it genuinely clarifies the argument.
3. Re-read each sentence and remove words that do not perform useful work.

How the TARA AI Grader Analyses a Response

ExamPrepIQ organises its independent practice feedback into four dimensions. These are feedback categories used by the platform; they should not be mistaken for an official UAT-UK marking rubric.

1. Interpretation & Meaning

This dimension considers whether the response merely paraphrases the statement or examines what its central terms and assumptions mean.

The 283-word response began with a clear explanation: excessive reliance on tools such as ChatGPT could cause students to lose the ability to think deeply for themselves. The grader credited this as a competent restatement.

However, the feedback noted that the response accepted important phrases at face value. It did not explore questions such as:

• What counts as “increasing reliance”?
• Does “ultimately diminish” suggest an inevitable outcome?
• What exactly is included within “critical thinking”?
• Is all use of AI equivalent, or do different uses have different effects?

Response B went further. It described critical thinking as a skill developed through “productive intellectual struggle” and explained the proposed causal mechanism: when AI bypasses that struggle, the relevant cognitive processes may be outsourced.

That distinction explains the 90/100 versus 40/100 interpretation scores. One response restated the claim accurately; the other unpacked its logic.

2. Counter-argument & Reasoning

The shorter response contained several relevant counterpoints. It argued that AI could find research material, summarise long chapters, proofread errors, act as a personal tutor and free students to focus on larger ideas.

These are valid ideas. The problem was development.

The grader observed that the essay often moved from one assertion to another without explaining the mechanism connecting AI use to stronger critical thinking. An example cannot replace the reasoning it is meant to support.

Consider the response's calculator analogy: AI might handle repetitive work in the same way that calculators handle routine calculation. The grader identified this as a promising idea, but one mentioned too briefly. A stronger paragraph could have explained that automating lower-order tasks may release attention for synthesis, evaluation and problem-solving.

The 476-word response did exactly that. It organised its counter-argument around AI as a “cognitive scaffold,” distinguished lower-order from higher-order work and connected concrete examples to a clear line of reasoning.

Its counter-argument therefore received 92/100, compared with 40/100 for the later attempt.

Do not stop after stating what an example shows. Explain why it supports your conclusion.

A useful paragraph sequence is claim → mechanism → example → link back to the statement.

3. Evaluation & Judgement

The shorter essay did state a qualified position:

“I agree with the statement to a moderate extent.”

That directly answered the extent question and avoided an overly simple yes-or-no conclusion. Its final sentence also made a useful distinction: the problem was not necessarily AI itself, but whether students used it to avoid thinking or to think more deeply.

The grader still found the judgement underdeveloped. The conclusion stated a sensible position, but it did not sufficiently weigh the competing arguments that came before it. The judgement felt asserted rather than earned.

Response B made its decisive condition much clearer:

“The tool itself does not dictate cognitive decline; the pedagogical design does.”

This sentence resolved the tension between the two sides of the essay. It identified the factor that determined the writer's final judgement: how AI is integrated into learning.

That is why its evaluation received 90/100.

A strong TARA conclusion should therefore do more than repeat the introduction. It should explain which argument is more persuasive, why it is stronger and what condition changes the extent of the writer's agreement.

4. Organisation & Expression

Response A's strongest dimension was Organisation & Expression, at 72/100.

The grader recognised that the response was easy to follow, divided logically between explanation, counter-argument and evaluation, clearly written and direct in its phrasing.

This is important because the overall 46/100 did not cause the system to dismiss everything about the essay. It preserved the distinction between strong presentation and weak development.

The feedback also supplied sentence-level revisions. It showed how two related sentences about intellectual struggle and instant answers could be joined to make the causal relationship sharper. It similarly suggested a more concise version of the sentence distinguishing AI as a starting point for research from AI as a final answer.

Response B received 95/100 in this dimension. Its structure was disciplined, its analytical language was precise and almost every sentence performed a clear argumentative function. Even here, the grader did not simply declare the response perfect: it suggested small edits to improve concision and transitions.

Why Did the 283-Word Attempt Receive 46/100?

The shorter response did several things correctly:

• It answered all three parts.
• It took a qualified position.
• It used relevant examples.
• It followed a logical structure.
• It communicated its ideas clearly.

Its central weakness was that these ideas often remained at outline level.

The overall feedback described the response as closer to a plan or summary of an argument than a fully developed argument. Because the interpretation, counter-argument and evaluation were all too brief to explore their ideas in depth, each of those three dimensions was capped at 40/100.

The grader's most useful diagnostic question was:

“Have I developed each of my points with reasoning and examples, or have I just listed them?”

That question gives the student a concrete focus for the next practice attempt.

Was the Lower Score Simply Caused by Writing Fewer Words?

No. The comparison should not be reduced to “476 words good, 283 words bad.”

The official limit is 750 words, not a target that every candidate must reach. A longer response can still be repetitive, unclear or poorly reasoned. The 476-word attempt performed strongly because its additional words were used to define terms, explain mechanisms, develop examples and weigh the competing positions.

For the specific 283-word response, ExamPrepIQ suggested aiming for at least 450 words in the next practice attempt and noted that approximately 500–600 words could provide room for fuller development. That was personalised feedback about this essay, not an official TARA minimum.

The better principle is:

Write enough to complete the reasoning, but make every sentence earn its place.

What the Stronger Attempt Did Differently

It defined the central concept

Instead of assuming that “critical thinking” was self-explanatory, it linked the concept to productive intellectual struggle, ambiguity and independent evaluation.

It explained cause and effect

It showed how passive AI use might outsource cognitive work rather than merely stating that AI could be harmful.

It built the counter-case around one central idea

Describing AI as a cognitive scaffold gave the counter-argument a clear organising principle.

It developed examples

The calculator analogy, research example and debate-partner example were connected to higher-order reasoning rather than presented as a list.

It made a qualified judgement

The conclusion identified pedagogical design as the condition determining whether AI weakens or strengthens critical thinking.

It remained concise

At 476 words, the response stayed well below the official 750-word limit while still providing enough space for development.

How to Use TARA Writing Feedback Effectively

An indicative TARA essay score is most useful when it changes what you do next.

Step 1: Check task completion

Confirm that your response explains the statement, gives a genuine reasoned argument against it and states and justifies the extent of your agreement.

Step 2: Identify the weakest reasoning stage

Do not revise the entire essay blindly. Decide whether the main issue is interpretation, counter-argument, evaluation or expression.

In the 283-word example, organisation was relatively strong. The priority was developing the reasoning inside that structure.

Step 3: Expand mechanisms, not lists

For each important claim, ask:

• Why is this true?
• How would it work in practice?
• What does the example demonstrate?
• How does it affect the original statement?

Step 4: Make the conclusion weigh both sides

Use explicit judgement language such as “The counter-argument is more persuasive because…”, “This risk is significant, but it is outweighed by…” or “The deciding factor is…”.

Step 5: Practise again under the 40-minute limit

Feedback becomes valuable when it informs another timed response. Repeating the task helps reveal whether the weakness has genuinely improved or merely disappeared from one untimed draft.

Follow our broader TARA preparation plan to combine writing practice with Critical Thinking, Problem Solving and full timed simulations.

What Useful TARA Essay Feedback Should Contain

A single number cannot explain how to improve. Effective TARA writing feedback should show:

• whether all parts of the task were addressed;
• separate strengths and weaknesses;
• evidence taken from the response;
• the reasoning behind each diagnosis;
• missed opportunities;
• specific action items;
• sentence-level improvements where appropriate; and
• one clear priority for the next attempt.

That is the role of ExamPrepIQ's TARA AI grader. It does not issue an official result. It provides a structured practice diagnosis that students can test in their next essay.

Try ExamPrepIQ's TARA Writing practice to complete a timed response and receive independent, indicative feedback on your argument, judgement, structure and expression.

Frequently Asked Questions

Practise, Diagnose and Retest

The difference between these two attempts was not that one response knew a secret essay formula. Both understood the topic and answered the required parts.

The decisive difference was development.

One response listed several good ideas. The other defined its terms, explained the mechanisms behind its claims, built a coherent counter-argument and used a clear deciding factor to reach a qualified judgement.

That is the kind of distinction detailed TARA writing feedback should reveal.

Start a timed TARA Writing Task with ExamPrepIQ, review the evidence behind the indicative feedback and use the action items to plan your next response. For answers about registration, test centres and results, see the TARA FAQs.