Green AI 🌳 “We want to shift the balance towards the Green AI option — to ensure that any inspired undergraduate with a laptop has the opportunity to write high-quality papers that could be accepted at premier research conferences.” 👩🏻‍💻
The focus on SOTA has caused a dramatic increase in the cost of AI, leading to environmental tolls and inclusiveness issues. We advocate research on efficiency in addition to accuracy (#greenai). Work w/ and at
Green AI: "[Deep Learning] computations have a surprisingly large carbon footprint. [...] This position paper advocates a practical solution by making efficiency an evaluation criterion for research along-side accuracy and related measures"
💻🧠+🌍🌳 recent reads: Green AI vs Red AI and "Tackling Climate Change with Machine Learning"
Interesting read from around introducing efficiency as an evaluation criterion for your models. Something we should all be considering especially given training a model can produce more CO₂ than an a lifetime of car travel 😨 #GreenAI
Green AI - interesting new preprint via the brilliant
"We propose reporting the financial cost or 'price tag' of developing, training, and running [machine learning] models" — say researchers should disclose how much computing power they use to encourage greener AI
8/ Metrics. Let's start measuring, reporting and comparing training times in papers. Beyond that, is there any interest in a semi-automated competition to create "Green" models whose innovation does not merely rely on more data or compute? 🌱
Green AI. (arXiv:1907.10597v1 []) #NLProc
"Green AI refers to AI research that yields novel results without increasing computational cost, and ideally reducing it" ⇢ new essay from and
While I'm at it: If you're worried about emissions in research, worry about travel. Our group does plenty of energy-intensive modeling work ("Red AI", ), and NYC doesn't really use renewable electricity...