Nonetheless, it is a serious problem when summaries distort news headlines, and extreme cases where this happens are unfortunately unavoidable. Apple can’t just fix these summaries with a software update. The only answers are to either help users understand the downsides of the technology so they can make more informed decisions, or to remove or disable the feature altogether. Apple apparently opts for the former.
We’re oversimplifying a bit here, but in general, LLMs like those used for Apple’s notification summaries work by predicting portions of words based on what comes before and are not capable of truly understanding the content which they summarize.
Additionally, it is known that these predictions are not always accurate, and incorrect results occur many times per 100 or 1,000 outputs. As models are trained and improvements are made, the error percentage may be reduced, but it never reaches zero when countless summaries are produced every day.
Deploying this technology widely without users (or even the BBC, it seems) really understanding how it works is risky at best, whether with the iPhone’s news headline summaries in notifications or Google AI summaries at the top of search engine results pages. Even if the vast majority of summaries are completely accurate, there will always be users who see inaccurate information.
These summaries are read by so many millions of people that the magnitude of the errors will always be a problem, regardless of the relative accuracy of the models.
We wrote at length a few weeks ago about how the Apple Intelligence rollout seemed rushed, contrary to Apple’s usual focus on quality and user experience. However, with current technology, Apple could not have made any improvements to this feature to achieve a zero percent error rate with these notification summaries.
We’ll see how well Apple makes its users understand that the summaries may be wrong, but getting all iPhone users to truly understand how and why the feature works the way it does would be a tall order.
