Site’s Top 10 Stories in 2021 Include Deep Learning Diminishing Returns, Global Issues
IEEE, the world’s largest technical professional organization for the advancement of technology, reviewed its most popular stories from 2021 in an article by writer Eliza Strickland.
The story found that several of their most important articles were on the flaws of deep learning. The number one ranked story was about how deep learning has diminishing returns based on the computational and energy costs of training deep learning systems. According to Strickland, several came from Spectrum’s October 2021 special issue on AI, The Great AI Reckoning.
Here are the highlights:
- Deep Learning’s Diminishing Returns
- 15 Graphs You Need to See to Understand AI in 2021
- How DeepMind Is Reinventing the Robot
- The Turbulent Past and Uncertain Future of Artificial Intelligence
- Andrew Ng X-Rays the AI Hype
- OpenAI’s GPT-3 Speaks! (Kindly Disregard Toxic Language)
- Fast, Efficient Neural Networks Copy Dragonfly Brains
- Deep Learning Isn’t Deep Enough Unless It Copies From the Brain
- The Algorithms That Make Instacart Roll
- 7 Revealing Ways AIs Fail
Deep learning requires tremendous resources to work, and the story explores the downside of it and the potential upsides.
“They analyzed the improvements of image classifiers and found that ‘to halve the error rate, you can expect to need more than 500 times the computational resources.’ They wrote: ‘Faced with skyrocketing costs, researchers will either have to come up with more efficient ways to solve these problems, or they will abandon working on these problems and progress will languish.’ “
The story provides links to a treasure trove of valuable insights into AI in 2021.
read more at spectrum.ieee.org