AI Doomers vs Builders: The Open Letter Era

Archive note: this is a backdated post, written years later while rebuilding this site. It’s dated to the moment it covers, but the hindsight is real.

On May 1, Geoffrey Hinton — the “godfather of deep learning,” whose work underpins everything I’ve been writing about since November — left Google so he could, in his words, speak freely about the risks of the technology he spent fifty years building. This follows March’s open letter from the Future of Life Institute, signed by thousands including Wozniak and Musk, calling for a six-month pause on training systems beyond GPT-4.

Nobody paused. Nobody was ever going to pause. But the spectacle of the field’s founders publicly worrying about their creation deserves more than the two lazy responses dominating my feeds — “doomer cult” and “they’re just hyping their product.” A working developer’s attempt at holding it honestly:

What I notice from the cheap seats

The worriers are not the usual suspects. When the loudest warnings came from philosophers and sci-fi, dismissal was easy. Hinton is the man whose students built these systems. When the person who designed the engine says he’s started worrying about the brakes, “he doesn’t understand the technology” is not an available rebuttal. That doesn’t make him right — domain founders have been wrong about their fields’ futures forever — but it changes the prior.
Both camps argue past the actual middle. The debate stages itself as extinction risk versus full speed ahead, while the territory I can personally verify sits in between: job-ladder disruption, industrial-scale misinformation, systems shipped at panic speed with theoretical fences, concentration of capability in three companies. Every item on that list is observable now and gets less airtime than the robot apocalypse. The speculative risk debate is, conveniently for everyone, unfalsifiable; the mundane harms have tickets filed.
The pause letter’s real content was its impossibility. A six-month voluntary halt in a race with billions at stake and national rivalries attached? The letter wasn’t a plan; it was a flare — the people building this do not believe anyone is steering. That part landed regardless of signatures.
“Trust us” is doing enormous work. The same labs warning about catastrophic capability are racing to build it, on the logic that someone will, so better the careful ones. Maybe. I’ve heard that argument before from every industry that later got regulated for good reason. Self-assessed carefulness is not a control.

Where a PH dev actually stands

From here, the debate has a colonial echo nobody in San Francisco mentions: the risks are being weighed, and the rules drafted, entirely by the countries that own the models. The Philippines — whose BPO industry employs over a million people doing exactly the language-and-process work these systems are best at — gets the disruption on schedule and the policy table never. Whatever “alignment” means, aligning the benefits with the people absorbing the labor shock isn’t on any letter I’ve seen.

My position, banked

For the record, written mid-debate: I use these tools daily and they’re the most useful things to enter my workflow in a decade. I don’t lose sleep over paperclip maximizers. I do think shipping civilization-scale interventions at competitive speed with no liability framework is how every historical “how did we let that happen” started. Both can be true. The builders are right that doom is speculative; the doomers are right that nobody is steering.

History says we’ll get the regulation after the incident, not before. I’d be delighted to update this post wrong.