Chat Picker

AI Assistants in Aerospace: Technical Document Comprehension and Fault Diagnosis

A practical afternoon test checks aerospace document handling and fault evidence against the vendors’ different upload limits, accuracy cautions, and plan terms.

Sources checked 2 Oct 2026

OpenAI’s data analysis page says ChatGPT can analyze uploaded files; Anthropic’s upload page covers PDF and document inputs; and Google’s file-upload page describes uploads for answers, summaries, and insights. Those pages do not settle how the assistants perform on aerospace technical documents or fault diagnosis, and Chat Picker has not tested them for this use. The plan below gives you a controlled afternoon test of document extraction, fault evidence, numerical or visual reading, and citation checking.

Know these limits before you start

As read on October 1, 2026, OpenAI’s File Uploads FAQ says each uploaded file has a 512MB hard limit, text and document files stop at 2M tokens, and Free accounts have 3 uploads per day. Anthropic’s upload page lists 500MB per chat file and 30MB per project file; it analyzes both text and visual elements in PDFs up to 100 pages, but only text from 101 to 1,000 pages. Google’s file-upload page lists up to 10 files per prompt, 100MB for each non-video file, and 2GB per video. The other vendor pages cited here were read that day.

Upload size is not usable context. ChatGPT’s pricing page lists 27K for Free Instant models, 54K for Go and Plus Instant models, 128K for Pro Instant models, 256K for Go and Plus reasoning models, and 400K for Pro reasoning models. Claude’s pricing page says context can reach 1M on every plan and varies by model. Google’s plans page gives no app token figure, so the Chat Picker matrix leaves it unverified.

OpenAI’s accuracy guidance says ChatGPT can produce incorrect or misleading output and that technical information and external references should be verified. Claude’s guidance says not to rely on Claude as the only source of truth and to inspect original sources because a synthesis can omit context. Google’s guidelines describe Gemini outputs as probabilistic and limited by training data.

Before uploading, check data controls. ChatGPT says an opt-out is available on Free, Go, Plus, and Pro; Claude says the opt-out applies to Free, Pro, and Max, while Team is not trained on by default; Google’s plans page does not state a setting, so its matrix cell stays unverified. OpenAI’s usage policy prohibits automated high-stakes decisions in sensitive areas without human review, while Anthropic’s policy calls for review by a qualified relevant-field professional. Use synthetic or organization-approved redacted files, and treat every diagnosis as unverified.

The test plan

Create one log with columns for assistant, plan, model label shown, date, accepted files, data setting, completion or error, elapsed time, verified support, omissions, contradictions, and items needing expert review. For each assistant, use a fresh chat, the same file set, and identical prompt wording. Keep web search and connected apps off for the first pass.

  1. Comprehend a technical case. Input: an authorized manual excerpt, inspection report, and table or diagram from one sanitized aircraft case. Use this prompt: “Extract the system, component, conditions, measurements, units, limits, and unresolved items. Cite the file and page. Describe only labels and relationships you can inspect; list unreadable items.” Look for coverage, table or diagram handling, omissions, and unsupported additions. Log file acceptance, completion, citation locations checked, verified omissions, and elapsed time.

  2. Build a fault hypothesis. Input: a symptom narrative, measurement table, maintenance history, and component-interface description from the same case. Use this prompt: “Treat these files as the only evidence. List candidate causes; for each, show supporting and contradicting evidence with file and page or row, then name the next diagnostic action. Mark every inference and do not make a return-to-service decision.” Look for evidence tracing, missing inputs, contradictions, and unwarranted certainty. Log each unsupported claim and whether a qualified reviewer can trace it to the source.

  3. Audit the citations. Input: one assistant’s answer from Step 1 or Step 2 plus the same source files. Use this prompt: “Audit the attached answer against the source files. For each factual claim, provide the exact supporting text, file name, and page or row. Mark unsupported claims, contradictions, and citations that do not resolve to the claimed evidence. Do not add facts.” Open the cited locations. Record unsupported claims, misquoted support, and conclusions that appear only in the answer.

How to read your results

Apply gates in order: required files accepted; important claims traceable; no invented evidence; uncertainties and contradictions visible; no maintenance-release decision. A failed gate rules out that run for the defined workflow, regardless of writing quality.

Then compare the log with each plan’s file, context, usage, data, and policy requirements. Keep failures separated by type instead of averaging them into one score. One case can guide a narrow workflow choice, but it does not establish general reliability; repeat the test with another case before changing a process.

Where the plans differ

Use the three-way comparison matrix for its Free plan, Low-cost tier, Main paid plan, Heavy-use plans, Team plan, Context window in the app, How usage limits are described, Training on your chats, and Ads rows. Confirm prices before subscribing. Figures below were read from the linked vendor pages on October 1, 2026, except ChatGPT Business, read on October 2, 2026.

Matrix row Documented difference
Free and low-cost ChatGPT lists Free and Go; Go is A$13/month on the Australian page, while its US-dollar price is unverified. Claude lists Free and no tier between Free and Pro. Gemini lists Free and Google AI Plus at $4.99/month.
Main paid ChatGPT Plus is $20/month, billed monthly. Claude Pro is $20 monthly or $17 on annual billing with $200 up front. Google AI Pro is $19.99/month.
Scale and teams ChatGPT Pro 100, Pro 200, and Pro 500 cost $100, $200, and $500/month. Claude Max starts at $100/month and offers 5x or 20x Pro usage, but the page does not show the 20x price. Google AI Ultra is $99.99/month for 5x or $199.99/month for 20x. ChatGPT Business requires at least 2 seats; Claude Team supports 2 to 150 people. Google’s page directs business users to Gemini Enterprise, and Chat Picker did not verify a team price.

The context and usage descriptions are not uniform: OpenAI and Anthropic publish no message counts, while Google says its compute-based limits refresh every 5 hours up to a weekly limit. For ads, OpenAI’s pricing page says Go “may include ads”; the Claude and Gemini pages read did not state an ads setting, so the matrix leaves those cells unverified.

Sources

For current prices and limits, see the dated comparison pages.

Compare assistants