3,580 karma · joined June 27, 2017
No you haven’t
The physics, however, don’t change, and neither does the task in front of us.
I just spent ten months finding a job, the longest and most difficult job search of my career. It was hard because the macro environment is terrible, but also because I insisted on finding a role working on clean energy hardware.
At a time when basically all VC funding, and in many ways the entire productive capacity of the economy, is shifting to AI, it would have been smarter in some respects to just pivot to AI. I sat out the crypto bubble too, in which some of my peers made a lot of money by working on projects that were either useless, or straightforward scams.
We crossed the 1.5 degree threshold because the money can’t help itself. The slow, steady work of transitioning away from fossil fuels is never going to be as profitable, as quickly, as the various forms of gambling and addiction that have driven the tech industry for most of my lifetime.
But you, individually, can make the choice to keep your eye on the ball. You can commit to working on this problem until it’s done, no matter what. You can choose to find the others to organize and prepare, so that the ideas closest at hand when the next crisis comes are yours.
It’s the best solution I’ve come up with so far, and in spite of the challenges, it’s still working.
did you though
State sponsored intelligence agencies can point a camera at an exterior window and reconstruct dialogue from the vibrations. This was a loose, hand-soldered ESP32. Settle down.
It's very easy to see how this goes — one side of the ledger is the cost savings you get from remote monitoring, remote control, automated cloud backups of access logs from employee badges, or whatever dumb reason this system was internet-connected in the first place.
On the other side of the ledger is the potential cost and probability of a shutdown from a cyberattack, whether that's from a belligerent state actor, a malfunctioning or maliciously deployed AI agent swarm, weird accelerationist hackers, or whatever, plus the cost of constantly defending and hardening systems against attack.
Chances are, the cost/benefit analysis will heavily favor just unplugging the thing from the Internet.
"Oh ho ho, unemployed? Why not spend the thousands of dollars you certainly have just lying around to dick around with chatbots, with all the spare time you have?"
If this is how you're thinking, you are not living in reality.
It's meant to assign agency and accountability where it actually lies instead of mystifying it with anthropomorphic language.
Failing to do so has real and harmful consequences, such as enabling OpenAI to escape accountability for clearly criminal behavior.
> many people are already focused on it, and it is therefore not an especially "neglected" area.
OK. Spend an afternoon reading about the current funding environment for decarbonization projects, vs the amount of capital available for AI, which is driving an increase in carbon emissions.
When I first encountered this argument back in the 20-teens, it didn't make sense, because there was effectively no forward progress on decarbonization. Much has changed since then, and the transition is moving, but it's moving in fits and starts, in punctuated bursts driven by reactions to long-predicted disasters actually showing up at the doorsteps of people with money (see, for example, the explosion in climate tech funding following San Francisco's "orange sky day" in 2020).
If we had an AI-data-center-boom-sized pile of cash to set on fire, we'd have completed the clean energy transition already. What's happening instead is that any hard tech company related to reducing emissions that hasn't already pivoted to some kind of AI-related marketing message is struggling to stay afloat, while losses at the likes of OpenAI and Anthropic are expected to accelerate, and everyone seems fine with that.
All the EA language around climate change is CYA bullshit. If this was true:
> Working on this issue seems to be among the best ways of improving the long-term future we know of..."
Then the above conditions would not obtain.
No, it isn't.
A human prompted an LLM to build a software simulation environment for hardware design, enabling an LLM, when prompted by a human, to optimize hardware designs against constraints in the simulation.
Lots of us working in clean energy read this book, and everything that’s happened since it was published makes the projections here useless
Thankfully, energy storage is being deployed worldwide, so this is a moot point, and renewables can and will replace fossil fuel electricity generation.