Get the latest tech news

Predictions from the METR AI scaling graph are based on a flawed premise

Just a because a graph is intriguing doesn’t mean that it means very much

METR’s blog and its tweets (perhaps written by publicists or by generative AI rather than by scientists), however, dropped the careful qualifications and nuance of the technical paper and made claims that go far beyond anything that the study actually supports. The task is to implement functions to process payments and avoid duplicate transactions when they are coming in asynchronously from different time zones and currencies. The complete problem specification is 3000 words long – well worth looking at to get a sense of what is really involved in a moderately complex software engineering undertaking.

Get the Android app