YOUR PROJECT OUTCOME
An AI output review sheet covering claims, original sources, numerical checks, limitations, alternatives and a reasoned final decision. Learn when to suspend a conclusion and seek specialist verification.
Who it is for
Branding, marketing, real estate and advisory professionals, entrepreneurs and team members using AI for research, presentations, content and operational decisions.
What you need to start
No programming skills required. A recommended foundation before Raffaella Giove’s other programmes. Human judgement needs evidence, alternatives and checks: reviewers can also be wrong.
The curriculum
01What hallucinations are · 8 min
+
Fabricated or unsupported claims, nonexistent sources and contradictions. Distinguish these from declared creative content, outdated information and calculation errors. Outcome: classify the problem before correcting it.
02Plausibility is not evidence · 8 min
+
Confident tone, persuasive explanations, apparent citations and agreement with the user. An elaborate explanation may contain false premises. Outcome: separate convincing presentation from factual support.
03Human reviewer bias · 10 min
+
Confirmation bias, anchoring on the first answer and excessive trust in automation. Make an independent initial assessment and seek contrary evidence. Outcome: recognise a blind spot in your own review.
04Reasoning about data and alternatives · 10 min
+
Facts, assumptions and conclusions; sample and market; correlation and causation; denominators and uncertainty. Compare three possible explanations. Outcome: correct a conclusion that goes beyond the data.
05Independent verification · 12 min
+
Open the original source, locate the relevant passage and check its date and scope; recalculate using appropriate tools and independent sources. Asking the same system “are you sure?” is not independent verification. Outcome: produce a reproducible verification record.
06Decisions and accountability · 12 min
+
Cases involving market research, a property brochure and a marketing budget. Use, correct or suspend the output, explain the decision and assign a reviewer. Outcome: deliver a corrected version with explicit limitations.
The final project
A planned 45-minute workshop on three simulated cases: market research, a property presentation and an acquisition budget. For each case, isolate the claim, classify the error, cite the evidence, correct the text and justify the decision. Exercise data is fictional and does not describe real clients or properties.
Planned assessment criteria
A planned 15-minute assessment with ten questions on errors, sources and valid conclusions; a proposed pass threshold of eight correct answers, subject to piloting. The supported pathway also plans an individually assessed practical audit, under terms to be defined. No method guarantees error-free output or the automatic superiority of human judgement.
Tools and materials
- AI output review sheet
- Original-source and claim register
- Spreadsheet for numerical checks
- Protocol: define, isolate, verify, challenge, decide
Listed tools are examples, not partnerships. Access requirements and possible additional costs will be specified before enrolment.
← Back to the course catalogue