Making sense of AI outputs without panic, paranoia, or blind trust

Live on 22nd September 2026 2-3:30pm (UK Time)

Once people stop being afraid of AI, the next challenge is judgement.

This session focuses on how to read, question, and evaluate AI-generated content using the skills librarians already have — just applied in new ways. It demystifies why AI can sound so confident while still being wrong, and gives practical techniques for spotting problems early.

Rather than positioning AI as something to “catch out”, this session reframes evaluation as a calm, professional literacy skill — the same judgement librarians have always brought to information, now extended into AI contexts.

What this session covers

Spotting AI-generated content
- Common patterns, tone, and signals (and why none are foolproof)
- When it matters to know something is AI-generated — and when it doesn’t
Why AI gets things wrong
- Hallucinations explained in plain language
- Bias, gaps, and over-confidence in outputs
- Why “it sounds right” is not the same as “it is right”
Confidence tricks and credibility signals
- How AI uses structure, fluency, and certainty to appear authoritative
- Why traditional trust cues (polish, references, clarity) can mislead
Practical evaluation techniques
- Simple checks librarians can use themselves
- Clear, teachable techniques that can be shared with users, students, and colleagues
- Knowing when AI is “good enough” and when it absolutely isn’t
Translating information literacy into AI
- Mapping familiar IL concepts (authority, accuracy, context, purpose) onto AI outputs
- Helping others understand how to think, not just what to trust

Who this session is for

Librarians who are being asked questions about AI accuracy and reliability
Staff supporting students, researchers, or the public with AI-generated content
Anyone worried about misinformation, but tired of alarmist takes
Professionals who want to teach good judgement without needing technical depth

Who it’s not for

People looking for detection tools or “gotcha” tactics
Those wanting technical or coding-level explanations
Anyone expecting AI to be either perfectly reliable or completely useless

Outcome

Participants leave able to say:

“I know how to assess AI outputs calmly and clearly — and I can explain that judgement to others.”

Join a Live Empower Session