Are you concerned about AI power consumption? Artificial intelligence is an electricity hog. Google says its total greenhouse gas emissions climbed nearly 50% over five years, mostly due to electricity that powers AI data centers.

On the topic of electrical usage for AI - note that we have the technology to make nearly infinite amounts of electricity without meaningful carbon emissions. Burning fossil fuels to make electricity is a choice, not a mandate or necessity, at least in developed countries where massive AI datacenters are likely to be hosted.
 
1) California now has 100+ days this year where 100% of electrical needs were met with wind, water, solar, and grid battery
 
2) Washington State has the cheapest electricity on the planet entirely because of our hydroelectric capacity. It is not a coincidence that towns like Quincy, WA host staggeringly large datacenters. Zero emissions except for the occasional diesel backup generator ops check. Hydroelectric power is near-perfectly reliable so the generators are never really used much in practice.
 
3) Fission reactors produce almost no pollution whatsoever, we have hundreds of years of proven fuel reserves, and the technology (incl. safety) has improved considerably. 

...and even if that "net carbon positive" were not directly true, the argument still holds: dramatically accelerating research in alternative energy, decarbonization, materials science, energy efficiency, battery chemistry, grid management, automation, transportation efficiency etc. has an indirect (harder to quantify) benefit that should not be ignored. This isn't even considering the impact of accelerated drug discovery, improved equity in education, quality of life and other non-energy-related matters that still likely pass the "benefits outweighs costs" argument.  
 
And in the end, we are talking about a small % of the total grid load here - at a time when the grid is becoming greener rapidly due to the mass deployment of solar in particular (+ grid battery) - look to California's progress recently - it is truly astounding and the whole world will follow because we've now reached the turning point from a cost standpoint. Solar - not coal - is now the cheapest way to make electricity in most scenarios. The math keeps getting better, year after year. There is no end in sight to this progression.
 
The challenge: Shifting thinking from a scarcity mindset to an abundance mindset. AI is the key enabling technology that accelerates that abundance.  
 

I am indeed concerned. I think that over the next 10 years, increases in renewable energy sources will be neck and neck with India and Africa energy demand. Population growth and population decreases will also augment the regional shortages and necessarily drive energy trading across geographies. But the trading landscape is becoming more stressed at the same time, due to political crises and conflicts that I believe are fueled by other forms of shortages (water, petroleum, minerals).
AI may indeed helps us stay away from the crest of a wave by helping us with breakthrough technologies that either make compute more efficient or help us accelerate innovation. I am not aware of any academic or industrial success stories in this area and would be very happy to have someone contribute to this discussion any real world examples.
While I do think we will solve it, we are "dancing with the devil" and keeping probabilities for dystopian futures higher than we might if we were to focus more on sustainable food and energy systems. COP28 was a sharp reflection of the complexity of the discourse. 
In any case for the next year and the following 10 I count much more on human talent than on AI to address the closely related problems.

More concerned about the power consumption from a waste perspective.  If AI goes ungoverned, there will be massive amounts of wasted CPU/GPU and thus wasted energy and money.  Actual Energy consumption from truly value add AI services should be offset by the business benefits.