Hey,
I've only got two papers this month but they're both biggies. One is the latest UN report on overshooting (and getting back to) 1.5C of warming, and the other looks at the potential for AI to increase both fossil fuel extraction and renewable energy.
But indulge me in a non-science aside for a moment. Putting this together made me think about focus and when to go with "good enough" and when to give something your all. When work is busy (it usually is) it pushes me to rush through reading and summarizing science. Sometimes that's good b/c I can share more science each month with people who can use it.
But sometimes it's better to put way more effort in to polish the hell of something. Like my first time speaking to the board of directors at my last job took about 30 hours of prep to put together a really solid 15 minute presentation, which ended up being the right call to get the engagement I wanted.
Or recently I had a difficult solo coming up (in Gaelic!) so I put in a good bit of time to be able to sing it competently. An early mini-concert went fine, and people liked it. But I felt like I wasn't giving the song what it deserved, so I really thought about the lyrics and what they meant, rehearsed it another 30 or so times, and when I sang it in the full concert the reaction was completely different - several people said they were moved! I was really surprised. It still wasn't a flawless performance, but the extra time really paid off. But when I sing in nursing homes I try to emphasize quantity over quality - they get more joy out of more frequent imperfect performances than rare but high-quality ones.
So back to work - what do you need to call good enough and move on to the next thing? What one or two things might you want to really invest in to level up your impact? On to the science!
CLIMATE CHANGE:
The latest UN report shows options to get the earth back to 1.5C of warming after overshooting it (optimistically we could limit it to 1.8C). It's worth at least reading the 5 page executive summary, but a few things stood out to me. 1) every fraction of a degree of warming avoided (and reducing time in overshoot) decreases risks and damages, although some permanent negative impacts are unavoidable (especially for small island developing states). 2) The authors see that adaptation isn't competitive with mitigation, and that without adaptation we'll get stuck spending all our resources on disaster response rather than continuing on mitigation. 3) Cooling the earth will be much, much slower than warming it (at least 5* slower). 4) Carbon dioxide removal is essential, but without sharp reductions in gross emissions, we could fill up total geological storage reservoirs trying to maintain net-zero without even getting to net negative emissions and cooling. 5) Since protected areas are fixed but species distributions are moving, PAs will increasinly be mismatched with where species are and ecological adaptation will be needed.
ARTIFICIAL INTELLIGENCE & CLIMATE CHANGE
Alpine et al. 2026 is a tricky paper - it makes an important point even though none of the numbers mean much. In short they argue that 1) AI can lead to increased efficiency and production of both fossil fuels and renewables (by better targeting new wells and renewable siting, and making more optimal decisions about managing existing energy facilities), 2) those indirect emissions impacts of AI are far higher than the direct energy use of data centers, and that 3) on net we can expect AI to increase fossil fuel use more than it boosts renewables. This overall conclusion is likely correct despite being based on 2017 data (when the world was very different). The drop in renewable price and increase in energy storage means there's less room to boost them. For fossil fuels, the huge jump in price means that many marginal resources are being tapped already w/o AI (price has driven new wells already), but for big companies w/ lots of wells there are a lot of efficiency gains in adjusting operations of existing wells. They find that emissions from small gains in fossil fuel extraction outweigh the benefits of larger improvements in renewable (partly b/c 80% of global primary energy is fossil-based). Due to the kind of model they used and much things have changed since 2017, and the fact that they omitted methane and other non-CO2 gases, the quantitative estimates are safe to ignore. But the point that we should be thinking less about AI boosting renewables and data center direct footprint, and more about the ability of AI to make fossil fuel extraction economical for longer is a good one. See the "pathway identification" section on p8 for specific examples of what kinds of changes AI could drive to both fossil and renewable energy.
REFERENCES:
Alpine, W., Geldner, N., Alpine, H., & Chepeliev, M. G. (2026). AI-driven productivity gains enable more CO₂ emissions than they avoid in a global energy–economy model. Npj Climate Action, 5(1), 71. https://doi.org/10.1038/
United Nations Environment Programme. (2026). Limiting Overshoot: Navigating exceedance of 1.5°C and pathways towards return. United Nations Environment Programme. https://doi.org/10.59117/20.
Sincerely,
Jon
P.s. If you know someone who wants to sign up to receive these summaries, they can do so at http://subscribe.sciencejon.
No comments:
Post a Comment
Questions, comments, suggestions, and complaints welcome.