AI Limitations & Accuracy Notice

Artificial intelligence is a powerful tool, but it is not infallible. This notice explains the limitations of the AI features in Wizard Application™ (the “Service”) so you know what to expect and when to double-check its output.

Why AI output can be wrong

The AI models behind Wizard AI generate text and other output by predicting what is statistically likely to come next, based on patterns learned from large amounts of data. They do not look up verified facts the way a search engine or database does. This means output can be inaccurate, incomplete, out of date, or stated with more confidence than is warranted, even when it looks fluent and plausible.

Not professional advice

AI-generated content does not constitute legal, financial, employment, tax, or other professional advice. You should seek independent professional advice before relying on it for any significant decision. See our general Disclaimer for more.

Recruitment AI is decision support, not a decision-maker

Where AI is used to help with recruitment, such as candidate screening, candidate-to-role matching, or CV parsing (available on our Professional and Enterprise plans), it is designed to support your team’s judgement, not to replace it. Matches, scores, and summaries can be incomplete or miss relevant context, and must be reviewed by a person before any hiring decision is made. See our Human Oversight Statement for how we keep people in control of these decisions.

Bias in AI output

AI models can reflect biases present in the data they were trained on, even with safeguards in place. This is particularly relevant to recruitment features, which is why AI output there is treated as a starting point for human review rather than a final judgement, and why hiring decisions remain the responsibility of your team under the equality and anti-discrimination legislation that applies where you operate, for example:

• The Equality Act 2010 in the United Kingdom;
• The Fair Work Act 2009 and the Australian Human Rights Commission Act 1986 in Australia;
• The Human Rights Act 1993 and the Employment Relations Act 2000 in New Zealand; and
• Title VII of the Civil Rights Act 1964, together with applicable state and local anti-discrimination laws, in the United States.

This list is not exhaustive: you remain responsible for complying with the equality and employment legislation applicable in your own jurisdiction.

Language and locale limitations

Wizard AI can work in many languages, since it is built on general-purpose AI models with broad multilingual capability. Quality can still vary between languages, particularly for less common languages or for content that mixes multiple languages within the same request, so you should apply extra care when reviewing output that is not in English.

Known limitation areas

• Long or highly technical documents may be summarised in a way that omits detail.
• Search and matching features rank by relevance, not certainty, and can surface imperfect matches.
• Generated images or text may not always match the intent of an ambiguous request.
• AI features are only as good as the information available to them at the time of the request.

Your responsibility to review

You are responsible for reviewing and verifying AI-generated content before you rely on it, send it externally, or use it to make a decision. AI suggestions are always editable, and nothing is saved or sent without your confirmation.

Ongoing improvement

We continue to test, monitor, and improve our AI features. If you encounter output that is inaccurate or otherwise unsatisfactory, please tell us using our AI Content Reporting Policy.

Contact us

If you have questions about the limitations of a particular AI feature, contact us at support@wizardapplication.com.