Watch Desk posted an update
Generative AI can produce convincing answers from familiar patterns even when the facts are wrong, argues Nicola Sfondrini, a PwC cloud infrastructure and architecture partner writing for Forbes Councils.
Why it mattersHer sharper point is about evidence: ten documents repeating one outdated claim may look like ten confirmations to an AI system, when they all trace back to the same source. Sfondrini’s proposed fix is to give AI information about its sources, including where a claim came from, how fresh it is and whether it is authoritative. Sensible advice, though not a demonstrated cure for hallucinations.
Discuss: Should organisations prioritise giving AI more information, or better records of where that information came from?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.