iiinova — where your digital self keeps evolving

rriv team @rrivteam.bsky.social · 27/09/2026

Anyway, nice talking to you, didn't mean to get so carried away but it's interesting. This is a cool tool if you haven't seen it, status line available for agentic if you use that for code stuff. ecologits.ai/latest/

ecologits.ai

EcoLogits

210

rriv team @rrivteam.bsky.social · 27/09/2026

Sure, but we are out of time. Are you aware of the number of renewable energy projects that have been shut down in America, at the same time as the carbon emissions rate is a large positive, while it needs to be a negative, and is a rate that still increasing.

200

rriv team @rrivteam.bsky.social · 27/09/2026

I agree. I think that would have been a great response to the OP actually. It's definitely unethical to use AI unethically. People might get somewhere if they can learn what that means.

100

rriv team @rrivteam.bsky.social · 27/09/2026

Studies have shown a range, like .5% to 3%. It's set to grow, yes? AI is currently driving the energy sector in America and keeping coal plants open, and adding gas. Importantly, this is NEW addition to global emissions, with emissions rate still growing aggressively year over year.

210

rriv team @rrivteam.bsky.social · 27/09/2026

Again, you would need numbers showing the total impact of, say, total worldwide video production pre-AI vs total worldwide video production with today's tools. Appeal to the exponential is almost a logical fallacy of its own these days.

200

rriv team @rrivteam.bsky.social · 27/09/2026

For instance, I do not need to call for a limit on what is allowed to evaluate the ethics of how a commons are used by particular individuals.

000

rriv team @rrivteam.bsky.social · 27/09/2026

In terms of resource use, people can self manage that when they have awareness of the impacts. This applies to every kind of consumption, not just AI. An _imposed_ limit is not something we would suggest. you can use AI ethically or unethically, yes?

100

rriv team @rrivteam.bsky.social · 27/09/2026

I get the original context. I hope to broaden it into something that actually makes sense to discuss. Both sides of the original discussion are too narrow. Our major critique on AI infra is not the tech itself, but the poorly planned speed of the built out, while amplifies the impact significantly.

100

rriv team @rrivteam.bsky.social · 27/09/2026

It's a fair starting point, but not a full argument. You need to calculate the total aggregate impact of the expected number of users for both usage patterns. Actually, I think you may be falling into the trap of Jevons Paradox, among other things.

200

rriv team @rrivteam.bsky.social · 27/09/2026

No I am not suggesting a limit in the amount of art you are allowed to create. I am suggesting that the per-work basis argument does not appropriate address what's actually important to the environment: the full aggregate impact.

100

rriv team @rrivteam.bsky.social · 27/09/2026

Energy usage reductions when generating a single image using AI vs Photoshop do not in any way show that "the environmental impact of generative AI is largely overblown". They are not logically linked statements, and I hope you eventually see why this is not a valid argument.

210

rriv team @rrivteam.bsky.social · 27/09/2026

Yeah for sure, but what I'm trying to get that the perspective that 'ai environmental impacts are overblown' being supported by 'diffusion generates an image using less power than a person drawing in photoshop' is not a well constructed argument and has tons of assumptions baked in.

100

rriv team @rrivteam.bsky.social · 27/09/2026

I'm also reading that on an RTX 4090, generating a high-quality image typically takes 2 to 5 seconds for standard models and 6 to 20 seconds for newer, more complex models. So the 1.5s math seems off?

110

rriv team @rrivteam.bsky.social · 27/09/2026

Just to add another dimension, research indicates that 70 percent of the total energy needed to make and operate a typical laptop over its lifespan is consumed during the manufacturing phase, rather than during daily use.

100

rriv team @rrivteam.bsky.social · 27/09/2026

And there was certainly a power plant plugged into the factories using ultraviolet lithography to make those chips my friend. And running the pumps and filters needed to produce the ultrapure water in the silicon wafer supply chain. And many other energy intensive processes.

210

rriv team @rrivteam.bsky.social · 27/09/2026

Also are we talking creating "art" or just comparing how fast an a single "image" can be created? I have use a post-it note w/ a pen and make a funny comic in about 8 seconds w/out a power plant plugged into me.

110

rriv team @rrivteam.bsky.social · 27/09/2026

The multi-modal models integrate LLM w/ diffusion, we both know, sorry for abbreviating. Both need to be accounted for. Similarly, we have to bound the supply chain to make a LCA comparison. Additionally, data centers also consume idle power, like a laptop - that must be counted in both cases.

210

rriv team @rrivteam.bsky.social · 27/09/2026

This comparison is heavily skewed by 3 things: high assumptions about time needed to create in photoshop, lack of analyzing context size and number of turns against the LLM, and finally assuming that drawings must be created in a computer at all. You could be right, but it's barely an analysis.

210

rriv team @rrivteam.bsky.social · 27/09/2026

Hmm, how much energy? Have you quantified that? You mention photoshop, but what about art made with paper and pencil. Also 'generate a image' - do you mean a single turn is all art requires in your model? How much token context do you send? I think we need to see your math on this.

210

rriv team @rrivteam.bsky.social · 26/09/2026

Can you show the calculations you are using to produce this conclusion? How do you quantify energy use for drawing (brain + motor control) vs the multi-turn image generation iteration workflow. You must have assumptions about specificity of output? We'd be interested to see your reasoning.

200

rriv team @rrivteam.bsky.social · 26/09/2026

A reminder of simpler times - look at this old cover and remember how clever and humorous these scientists were being about fascinating and important environmental science. Absolute, life long favorite journal. ❤️

020

rriv team @rrivteam.bsky.social · 20/09/2026

Great article. The solutions are there, it's the mindset that's lacking.

000

rriv team @rrivteam.bsky.social · 19/09/2026

Fish, water, temperature, energy: the columbia river columbia.rriv.org.

020

rriv team @rrivteam.bsky.social · 11/09/2026

Checking in on an old project of mine with some new tools

030

Reposted by rriv team

Peter Gleick @petergleick.bsky.social · 09/09/2026

The US Court of Appeals for the 5th Circuit has ruled there is no constitutional right to clean #water. The UN has declared a human right to water, but that has no sway here. If there's no "constitutional" right, federal & state laws must provide that protection. www.theguardian.com/us-news/2026...

theguardian.com

US federal court says no constitutional right to clean water in Mississippi case

Appeals court dismisses lawsuit from Jackson residents claiming that tainted drinking water violated their rights

66030

rriv team @rrivteam.bsky.social · 05/09/2026

The world is going need a whole lot of restoration after the current cycle of war and industrial intensification.

010

rriv team @rrivteam.bsky.social · 02/09/2026

To generalize it, AI is being used to spin the existing economy faster. The existing economy has some mis-design baked in; going faster accelerates its negative consequences further beyond the resiliency of ecosystems. This outcome is not an inevitable, but it is the one currently being chosen.

030

rriv team @rrivteam.bsky.social · 01/09/2026

A major problem with the AI data center build out is that its resource use is not even necessary for its own stated purposes. theaifootprint.com

110

rriv team @rrivteam.bsky.social · 01/09/2026

it's gonna sneak back in later.

000

rriv team @rrivteam.bsky.social · 28/08/2026

Wow there was a typo there, you are right again. I really have to have thank you for that, how embarrassing, fixed now!

000

rriv team @rrivteam.bsky.social · 28/08/2026

When everything runs faster, you tend to use more energy, which means waste accumulates faster and resources are gone sooner.

020

rriv team @rrivteam.bsky.social · 27/08/2026

oh thanks for the bug report, looks like bluesky's default doesn't append https to bare links. fix it up now. have a great day!

100

rriv team @rrivteam.bsky.social · 27/08/2026

I specifically DID read the article my friend, and offered my take on the issue to continue discussion. There is NOT currently a solution to the data center issues, which is why engineers need to slow down and fix it first. The technology WE build is for monitoring industry, one part of a solution.

100

rriv team @rrivteam.bsky.social · 27/08/2026

Deploying equipment into the ocean or space is vastly more complex than land sites - it comes w. all the same challenges and adds much harsher environmental conditions. There's no good solution right now and they know it. These efforts are more show than practical solutions. Slow it down folks.

100

rriv team @rrivteam.bsky.social · 27/08/2026

What's needed are regulations that require responsible, appropriately paced build out - at a minimum. We can go way beyond that and require responsible USE, that's what's needed for society.

000

rriv team @rrivteam.bsky.social · 27/08/2026

environmental purification agency

011

rriv team @rrivteam.bsky.social · 23/08/2026

chip fabs are using more and more recirculated water over the years, but it's far from a closed loop. water is downcycled for other uses in the fab, there is still a large consumption. either way energy, over build, and poor future planning are also features of this industry, we agree.

000

rriv team @rrivteam.bsky.social · 22/08/2026

The water used to run the data centers is one thing, the water used to make the soon to be obsolete chips themselves is way more.

100

rriv team @rrivteam.bsky.social · 20/08/2026

All creatures of this planet are in for a very hard time. We will have to fight hard for them, and for ourselves, as the machine world continues gaining strength. We will have to make sacrifices, and keep up with a vision.

030

rriv team @rrivteam.bsky.social · 19/08/2026

We are entering an age of ubiquitous robotics, and we must prepare to study these machines in the way we have studied ecosystems - because their presence will impact every corner of the globe. If we take this seriously, we may yet find a way to a sustainable world.

010

Reposted by rriv team

The Overhead Wire @theoverheadwire.com · 18/08/2026

You might not be following this because it's not about cities per say, but the administration just repealed the 2001 roadless rule in National Forests. They want to open up for more logging and more building which messes with everyone's drinking water. www.nytimes.com/2026/08/18/c...

nytimes.com

Trump Administration Moves to Allow Logging in Pristine National Forests

The move to repeal the ‘roadless rule’ would pave the way for road construction and logging in nearly 45 million acres of untouched back country.

46238

rriv team @rrivteam.bsky.social · 18/08/2026

It's going to be a terrible time for mother earth. At a minimum we need to document every step of the destruction and raise the alarm. The baseline is set to shift to an even more degenerated world ecosystem.

000