Its that time of year again: Pumpkin spice permeates the air, Thanksgiving flight prices are on the rise, and the annual State of AI report is fresh off the presses.
Spoiler alert: Transformers are in, semiconductors are hot, and AI + structural biology make a great couple.
Quick recap: Every year, Emerging Tech Brew digs into the report, which has been published by UK investors Nathan Benaich and Ian Hogarth since 2018. The authors 2020 predictions netted a ~68% accuracy rate: Nearly six out of eight became a reality, including a new structural-biology breakthrough from DeepMind and Nvidias failure to complete its Arm acquisition.
This year, the report covered top takeaways for research, talent, and politics. Well break down a few of them...
Biology: AI is taking over the structural-biology world, yielding new insights about both cellular machinery and protein shapeswhich could lead to not only drug discovery, but also better understanding of the human body and how diseases work.
Archi-tech-ture: Transformers, a popular architecture used to create models for natural language processing, are now being applied to many other machine learning use cases, like protein structure prediction and computer vision.
Large language models: The powerful and controversial AI tools, which underpin many of the services we use every day, are scaling up and out. And individual countries are looking to create their own dedicated LLMs.
AI talent: Academic organizations dont have enough compute resources to accomplish their projectsbut 88% of top AI faculty have received Big Tech funding, the researchers wrote.
AI safety: Although AI safety is making headlines, fewer than 50 researchers are working on it full-time at the largest AI labs, according to the reportmeaning there are a lot more people whose full-time job it is to build this tech rather than think about potential consequences.
Chips on chips: Semiconductors are hot, hot, hotand the global chip shortage is only increasing demand. Startups in the space are accelerating in a big way, and both countries and corporations are looking to make the semi supply chain more efficient.
Well see you back here next year to see how these sectors of AI move forwardand, of course, which of the reports predictions came true.Read More..