Dario Amodi, Five Months discussed on Software Engineering Daily


Applications of artificial intelligence are permeating our everyday lives we notice it in small ways improvements to speech recognition better quality products being recommended to us cheaper goods and services that have dropped in price because of more intelligent production but what can we quantitatively say about the rate at which article intelligence is improving how fast are models advancing do the different fields in artificial intelligence all advanced together or are they improving separately from each other in other words if the accuracy of a speech recognition model doubles does that mean that the accuracy of image recognition will also double it's hard to know the answer to these questions machine learning models trained today can consume three hundred thousand times the amount of compute that could be consumed in twenty twelve that's a nice statistic but it doesn't necessarily mean that models are three hundred thousand times better these training algorithms could just be less efficient than yesterday's models and therefore they're consuming more compute but we can observe from impure data that models tend to get better with more compute they also tend to get better with more data input how much better do they get do they scale linearly with the amount of data or the amount of compute well that varies from application to application varies from speech recognition to language translation we can't really say anything conclusively about all machine learning models improving because of some specific metric but models do seem to improve with more compute and more data dario amodi works at open ai where he leads the ai safety team in a post called ai and compute dario observed that the consumption of machine learning training runs is increasing exponentially doubling every three point five months.

Coming up next