And "Also consider whether non-proprietary, energy-efficient, or small-scale systems suffice for your task. If not, consider how preservation of the values articulated in this Declaration may be worth a delay in obtaining results."
Jul 21, 2026, 3:44 PM@timnitGebru@dair-community.social That's such an excellent statement and is kind of the reason I don't do much HPC anymore (and not professionally).
Large scale computing is always a devil's bargain, but I considered it worthwhile for computational chemistry and physics. But as I went from the academic side to the professional side, it became obvious how even in the 'traditional', pre-LLM computing world, only a small percentage of the computing resources were allocated to research. There was already a lot of waste (oil and gas simulations, etc). And in the current computing paradigm, it is so much worse...
From the declaration, I also really liked:
As mathematicians, and also as inhabitants of a shared world, we have a duty to care for other people and our environment.It's so simple, but also such a fresh breath of air after coming up against "we only care about the technology in this project" so often.
Jul 21, 2026, 3:51 PM@aud @timnitGebru this is so important, and points to the uneven allocation of *all* resources, not just computing resources.
The kind of budgeting that I do now in academia (social sciences) is laughable compared to the budgeting I did when I worked in tech.
And most of the big budgets and excessive resources are just waste...
@aud such a simple statement that I"m surprised to see from academics writing a declaration in a STEM field. What does that say smh.
@aud @timnitGebru there are some lovely counter-examples, though. I can offer one from my career: I briefly worked for a company that does analog semiconductor design (for cell phone front-end filters) in silico. Their work directly replaces (polluting) fabrication turns, which were previously the only way to evaluate performance for those filters. But when the compute got good enough and big enough, the simulations became comparably useful.