DTN 007: Generative AI has a Carbon Problem

‍

Generative AI’s Energy Use Makes Bitcoin Look Green

“While neither OpenAI nor Google, have said what the computing cost of their products is, third-party analysis by researchers estimates that the training of GPT-3, which ChatGPT is partly based on, consumed 1,287 MWh, and led to emissions of more than 550 tons of carbon dioxide equivalent—the same amount as a single person taking 550 roundtrips between New York and San Francisco.” (WIRED)

Japan Adopts Plan to Make Maximum Use of Nuclear Power

“The basic plan for green transformation marks a departure from Japan’s policy of reducing its dependence on nuclear energy decided after the 2011 meltdowns at Tokyo Electric Power Company Holdings Inc.’s tsunami-hit Fukushima No. 1 power plant. To the end, the government adopted green transformation legislation so that it can procure funds for its share of the envisioned investments, which is projected at some ¥20 trillion.” (The Japan News)

The Incredible Time Cost of Peer Review

“In 2020 alone, researchers spent 15,000 years worth of time reviewing articles. In the US, the time cost for peer review is equal to $1.5B, in China it is $600M, the UK is $400M.” (Research Integrity)

Underdog Technologies Gain Ground in Quantum-Computing Race

“For years, two leading approaches have enabled physicists to make progress partly by cramming devices with more and more qubits. One of those methods encodes qubits as currents running on superconducting loops. The other uses excited states of individual ions trapped in a vacuum by electromagnetic fields. But in the past two years, qubits that consist of single neutral atoms — as opposed to ions — and are held with ‘tweezers’ made of laser light have suddenly become competitive. And other techniques that are at an even earlier stage of development could yet catch up.” (Scientific American)