The engineer, who posted anonymously on X under the username voxium for fear of reprisal from his employer, called his new role at an undisclosed company "soul-sucking."
There’s another type, which is like me, who develop as a means to have a tool I need.
I wanted a better way to manage my cross-device libraries (desktop, laptop, NAS), and the multi-pane file managers didn’t work great for me, so instead I used Claude to build one that has native per-pane ssh support, streamed video and music preview clips, etc etc.
Capitalism is always gonna Capitalism, doesn’t matter what type of tools are at play.
If it weren’t for the implicit evils from mass scale LLMs that drain natural resources & are built off of mass theft, I would absolutely use them for little one-off low stakes things that rarely call for long term maintenance. But I cannot reconcile the relatively minor perks with supporting massively unethical & harmful trends & corporations.
Sadly, they exist irregardless of our little projects. Financially supporting them is one thing, and if your own ethics prevent you from using a weapon of the enemy, as it were, that’s fine, but abstention is not actually doing anything to oppose them.
The water consumption of any LLM compute time I’ve used is almost assuredly less than a single bath, and I’ve certainly skipped more baths in my life than I probably should have. :P
And more importantly, an hour of Netflix uses SIGNIFICANTLY more water than an hour of text LLM generation, so most people are not actually having the impacts on water consumption they likely think they are.
The support at this stage is usage statistics packaged as demand to legitimize further proliferation of data centers and further theft of IP to attempt to bump up the “knowledge” based fully off of mass quantities. This further creates an illusion of worth that they can exponentially sell to gullible upper management. In the end, once the grift reaches a terminal point, the only thing that will trickle down is the consequences.
We’re way beyond usage statistics. That’s for startups trying to get a Series A. OpenAI and Anthropic are playing at the Fortune-100 level already, influencing elected officials and making deals for government contracts. They’re not rolling up to the DoD with their MAU counts.
There’s another type, which is like me, who develop as a means to have a tool I need.
I wanted a better way to manage my cross-device libraries (desktop, laptop, NAS), and the multi-pane file managers didn’t work great for me, so instead I used Claude to build one that has native per-pane ssh support, streamed video and music preview clips, etc etc.
Capitalism is always gonna Capitalism, doesn’t matter what type of tools are at play.
If it weren’t for the implicit evils from mass scale LLMs that drain natural resources & are built off of mass theft, I would absolutely use them for little one-off low stakes things that rarely call for long term maintenance. But I cannot reconcile the relatively minor perks with supporting massively unethical & harmful trends & corporations.
Sadly, they exist irregardless of our little projects. Financially supporting them is one thing, and if your own ethics prevent you from using a weapon of the enemy, as it were, that’s fine, but abstention is not actually doing anything to oppose them.
The water consumption of any LLM compute time I’ve used is almost assuredly less than a single bath, and I’ve certainly skipped more baths in my life than I probably should have. :P
And more importantly, an hour of Netflix uses SIGNIFICANTLY more water than an hour of text LLM generation, so most people are not actually having the impacts on water consumption they likely think they are.
The support at this stage is usage statistics packaged as demand to legitimize further proliferation of data centers and further theft of IP to attempt to bump up the “knowledge” based fully off of mass quantities. This further creates an illusion of worth that they can exponentially sell to gullible upper management. In the end, once the grift reaches a terminal point, the only thing that will trickle down is the consequences.
We’re way beyond usage statistics. That’s for startups trying to get a Series A. OpenAI and Anthropic are playing at the Fortune-100 level already, influencing elected officials and making deals for government contracts. They’re not rolling up to the DoD with their MAU counts.