Why AI experts' jobs are always decades from being automated
allenai.substack.com
Experts say artificial general intelligence (AGI) will take around forty years to develop and won’t fully automate us for almost a century. At the same time, advancements like ChatGPT and DALLE-2 seem to send a different message. Silicon Valley and leading AI initiatives are anticipating this technology to spark radical change over the next decade – enough to redefine society from the ground up. So, why do AI researchers continue to predict AGI into the latter half of the century? And should we even take them seriously? 1. Experts don’t want to admit that AGI could be a decade away because it would make them irrelevant. AI experts' work is spread across the fields of machine learning and computational neuroscience. There are a plethora of research directions being explored. These include reinforcement learning, convolutional neural networks, training image classifiers, neurosymbolic modules, hybrid systems wanting to directly imprint inspirations from the brain… This is still just barely scratching the AI approach surface.
Why AI experts' jobs are always decades from being automated
Why AI experts' jobs are always decades from…
Why AI experts' jobs are always decades from being automated
Experts say artificial general intelligence (AGI) will take around forty years to develop and won’t fully automate us for almost a century. At the same time, advancements like ChatGPT and DALLE-2 seem to send a different message. Silicon Valley and leading AI initiatives are anticipating this technology to spark radical change over the next decade – enough to redefine society from the ground up. So, why do AI researchers continue to predict AGI into the latter half of the century? And should we even take them seriously? 1. Experts don’t want to admit that AGI could be a decade away because it would make them irrelevant. AI experts' work is spread across the fields of machine learning and computational neuroscience. There are a plethora of research directions being explored. These include reinforcement learning, convolutional neural networks, training image classifiers, neurosymbolic modules, hybrid systems wanting to directly imprint inspirations from the brain… This is still just barely scratching the AI approach surface.