Amazon has a surprising number of items that are called “I cannot fulfill that request” or something close to that. They have fewer after several websites highlighted the phenomenon yesterday and sent Amazon scrambling to take them down. That’s what happens when you use a large language model to name your products, and even worse, write your Amazon ads. While some of these ads may be the result of clueless translation attempts, far more of them are scammers, counterfeiters, and drop shippers who are generating so many ads that they cannot be bothered to actually look at them. And neither does Amazon.
Listen beautiful relax classics on our Youtube channel.
The ads that made it through without encountering any sort of quality control can be hilarious, like when the item name includes a mention of OpenAI’s rejection of trademark infringement. After the article was published, Amazon took down all the items that were screenshotted and more, but the same oddities are rampant in social media platforms.
We can laugh, but the real problems come when AI is somewhat supervised and therefore not so easy to detect. Read about the Amazon sellers letting AI run amok at Ars Technica. -via Fark
(Top image: Amazon via archive.org)
Source: neatorama