With all of the leaps and bounds forward when it comes to Artificial Intelligence (AI), there are occasionally sobering reminders that when it goes wrong, it can really cause some damage. The AI-generated news alerts feature rolled out with the newest iPhones is the most recent of these reminders. After widespread criticism from both users and news organisations that it was serving headline summaries containing regular errors, Apple have withdrawn the feature until the teething problems can be fixed.
In the UK, the BBC were one such group complaining about the accuracy of the feature, after an AI headline summary was served to some users, falsely claiming that Luigi Mangione, the person accused of shooting United Healthcare CEO Brian Thompson, had shot himself.
Not only did this feature send out false headline summaries, these summaries also looked like notifications coming from the news apps affected, which is a huge problem when it comes to the general public having its trust in news media eroded further.
It’s referred to as a ‘hallucination’ when AI models simply make something up, and as yet, there’s no real way to ensure that AI output is free of them, without being closely monitored by human eyes.
It makes sense that companies want to be at the head of the innovation queue when it comes to AI features, but there needs to be a balance to ensure that there isn’t widespread misinformation flying around which could do some serious damage.
Although they’ve removed the feature for now, Apple are planning on reintroducing it in a future update when they can be more confident in the information it provides.
What do you think? Should companies developing AI models slow down? Should there be more accountability in instances where the technology goes wrong? Come and join us over on LinkedIn to have your say.
You may also like: Is TikTok at risk of being banned in the UK?
