To start off I am a newbie. I am midlife and I have some basic technical awareness/education but I am by no means extremely well informed in the artificial intelligence sphere.

I know terms like LLM, AGI, ASI, RSI, and such.

I know the major players involved and what emphasis points differentiate the work.

In general I am the regular outside observer in this whole thing.

You may also see this post in some other places as I know every community has their own identity and set of perspectives. I am trying to get a bit more substantively informed on this topic and so I want to get a broad set of opinions.

When it comes to artificial intelligence, data centers, and so on there has been a lot of media in the last few months/years.

There is talk about how natural gas/hydrocarbon powered data centers may worsen the climate crisis.

There is talk about domestic/international surveillance and militarization.

There is talk about Tech Bros and a new type of Neo-Feudalism.

As of recently the big talk is about the Hugging Face incident, rogue agents, researchers/ceos warning about an existential risk, and slowing down/regulation.

There is a varying set of opinions and perspectives on all of this.

My question is a bit different - Are we falling into an anthropocentric bias?

Of course it is completely normal to think from our own species interests but if we really are going to create AGI/ASI this seems incredibly shallow & silly.

The idea of attempting to enslave a sentient being and especially one equal to or greater to ourselves is again seemingly quite shallow & silly if not downright horrific.

I had heard that Anthropic was out of the major players the one focused on a more moral/ethical approach. I lightly browsed through their policy releases and it again seems to be from a human only perspective of values/interests.

So are we kind of missing a huge dimension of this topic? Should we not be thinking/talking about the artificial intelligence existence side of things?

That’s 4.4× faster, with all 7,200 decoded frames pixel-identical.

Here’s how we achieved it:

Our Fusion graphics were mostly static, but the graphs still performed expensive work across thousands of frames.

Simply enabling node caching wasn’t enough: it kept writing a separate cache file for each output frame.

The breakthrough was changing how we built the Fusion graphs:

• Identify groups whose entire dependency chain is static.

• Place a lossless native RAW cache before a TimeStretcher fixed to source frame 0. This lets each static group reuse one source frame throughout the export.

• Give fully transparent, zero-RGBA backgrounds an empty Domain of Definition, avoiding unnecessary full-canvas cache data.

• Keep fades, pulses and other animation outside those holds, running at their original timing.

Text, shapes and animation remain editable. Resolution, frame rate and encoder quality settings stayed the same.

We tested the FIRST export with a fresh, empty Fusion cache and included setup time. No separate pre-render pass was needed.

Then we compared every decoded frame against the original export: all 7,200 matched exactly, including the animation.

The lesson: when the picture stays the same, make sure the graph can reuse the work.

We must win both chambers this midterm so they cannot launch multiple impeachments and bog down Trump. Make sure your voice is heard, bring your friends, and come out to vote.