Alan S. Morrison (@alansmorrison.bsky.social)

𝗜𝗳 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆 𝗼𝗳 𝗿𝗲𝘀𝘂𝗹𝘁𝘀 𝗶𝘀 𝘆𝗼𝘂𝗿 𝘁𝗼𝗽 𝗽𝗿𝗶𝗼𝗿𝗶𝘁𝘆 𝘄𝗵𝗲𝗻 𝘂𝘀𝗶𝗻𝗴 𝗮𝗻 𝗔𝗜 𝗰𝗵𝗮𝘁𝗯𝗼𝘁 𝗳𝗼𝗿 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵, 𝗿𝗲𝗮𝗱 𝘁𝗵𝗶𝘀 𝘀𝘂𝗿𝘃𝗲𝘆 𝗳𝗿𝗼𝗺 𝙏𝙝𝙚 𝙒𝙖𝙨𝙝𝙞𝙣𝙜𝙩𝙤𝙣 𝙋𝙤𝙨𝙩 𝘀𝘂𝗺𝗺𝗮𝗿𝗶𝘀𝗲𝗱 𝗯𝘆 Carly Martinetti www.linkedin.com/posts/prcarl... #AI

loading . . .

Washington Post tests AI tools for PR pros, reveals strengths and flaws | Carly Martinetti posted on the topic | LinkedIn The Washington Post just did all PR pros a solid by testing which AI tools deliver accurate information versus confident gobbledegook. They pitted the top 8 “AIs” against each other with 30 questions, and used human librarians armed with old-school Google Search as the control. Here’s what I think anyone in PR will find most revealing: 1. Google's AI Mode ranked highest when the librarians scored the accuracy of the AI answers with 60.2% accuracy. 2. ChatGPT-5 landed second at 55.1%, beating its predecessor ChatGPT-4 overall. It’s interesting to me that the older version performed better on bias and sourcing questions. 3. Perplexity did the best with image-based questions (it was the only one that could identify buttons on an iPhone), but all the tools struggled with hallucinations, confusion of people in the picture, or refusing to answer altogether. 4. Meta AI finished dead last at 33.7% accurate, and was... https://www.linkedin.com/posts/prcarly_the-washington-post-just-did-all-pr-pros-activity-7373362469961887744-XA1d/?utm_source=social_share_send&utm_medium=member_desktop_web&rcm=ACoAAAGqeWIBp0_UsIWj1ZEZCMrO-psf7LZT-GY