Knowledgebase

Articles

AI Ethics and Responsible Use: Everything That Matters, Briefly The section summarised.ON ACCURACYIt produces plausible text. Plausible and true are different,... AI and Confidential Business Information Protecting what matters commercially.WHAT COUNTSContracts and negotiations Financial records... AI and Fraud Current methods.WHAT HAS IMPROVED FOR ATTACKERSPhishing messages without the errors people were... AI and Misinformation Producing and encountering false content.WHAT HAS CHANGEDProducing convincing false content is... AI in Decisions About People Where the line sits.DECISIONS THAT SHOULD NOT BE AUTOMATEDHiring, promotion, discipline,... Accountability for AI Output Who is responsible.THE POSITIONThe person who sends, publishes or acts on output is responsible... Assessing AI Risk in Your Organisation What could go wrong.THE QUESTIONSWhat are we using it for? What would a wrong output cause? Would... Automation Bias Why people accept wrong answers.WHAT IT ISThe documented tendency to accept a system's output,... Copyright and AI-Generated Content An unsettled area.WHAT IS CONTESTEDWhether training on copyrighted material is permitted Whether... Data Protection and AI for Businesses Obligations when using these tools.WHAT TRIGGERS OBLIGATIONSProcessing personal data, which... Deepfakes and Synthetic Media Convincing fakes.WHAT IS NOW POSSIBLEVideo of a real person saying things they did not say Audio... Designing Review Into AI Workflows Making oversight real.WHAT TO DECIDE BEFORE BUILDINGWhat must be reviewed By whom With what... Disclosure: When to Say AI Was Used Transparency.WHEN DISCLOSURE IS REQUIREDWhere a regulator or professional body requires it Where... Handling Other People's Data Information that is not yours.WHAT COUNTSCustomer names, addresses and contact details Employee... How to Fact-Check AI Output Practical verification.WHAT ALWAYS NEEDS CHECKINGStatistics and figures Dates Names and titles... Keeping Human Skill and Judgement What to protect.THE RISKSkills erode when unused. Judgement weakens when not exercised.WHERE IT... Plagiarism and AI Originality.WHAT MODELS GENERALLY DO NOT DOReproduce training material verbatim, in most... Privacy When Using AI What happens to what you send.THE BASIC POSITIONText you send to an AI service leaves your... Representation and Whose Data Trained the System Who these systems serve well.THE PATTERNModels perform best on what they saw most.That means... Securing AI Use in Your Organisation Practical controls.THE ESSENTIALSApproved tools, named Individual accounts, not shared Strong... Stereotypes in AI Output What appears in generated content.WHAT HAPPENSGenerated text and images reflect patterns in... Talking to Customers About Your AI Use Transparency in business.WHEN TO TELL THEMWhere it affects them materially Where their data is... Testing for Fair Outcomes How to check.WHAT TO MEASUREOutcomes by group, not overall accuracy.Approval rates, error rates,... The Fabricated Citation Problem Why references cannot be trusted.WHAT HAPPENSAsk for sources and you receive citations: authors,... The Limitations Worth Remembering What has not been solved.RELIABLE FACTUAL ACCURACYNot solved. Models generate plausible... Training People to Use AI Responsibly What they need to know.THE FOUR THINGS THAT PREVENT MOST PROBLEMSIt produces plausible text, not... Understanding Hallucination The central limitation.WHAT IT ISConfident, fluent, false output. Invented citations, fabricated... Using AI Responsibly in Your Business What good practice looks like.WHAT RESPONSIBLE USE MEANSVerifying before publishing Not... Using AI With Other People's Work What you may process.WHAT IS GENERALLY FINESummarising a document for your own understanding... Verifying Numbers and Statistics Figures are the highest-risk output.WHY FIGURES ARE PARTICULARLY DANGEROUSThey carry authority... What AI Should Never Be Used For Clear boundaries.NEVER USE IT TODeceive anyone about what is real Impersonate a person Generate... When AI Is Confidently Wrong Recognising it.THE DIFFICULTYThere is no signal. Wrong output reads exactly like right... When to Stop Using AI for Something Recognising it is not working.THE SIGNSChecking takes as long as doing Errors keep reaching... Why AI Gets Things Wrong The mechanism behind the errors.WHAT THESE SYSTEMS DOGenerate text that is statistically likely... Why Human Oversight Matters The principle.WHAT AI CANNOT DOBe accountable Notice that it is wrong Understand consequences... Writing an AI Policy What to include.WHAT TO COVERApproved tools What may be sent, and what may never be What must be...

Back

Are you happy with your experience? Leave us a review on Trustpilot.