Anthropic CEO floats idea of giving AI a “quit job” button, sparking skepticism

Anthropic CEO Dario Amodei [raised a few eyebrows](https://www.reddit.com/r/OpenAI/comments/1j8sjcd/should_ai_have_a_i_quit_this_job_button_dario/) on Monday after suggesting that advanced AI models might someday be provided with the ability to push a “button” to quit tasks they might find unpleasant.

“So this is—this is another one of those topics that’s going to make me sound completely insane,” Amodei said during the interview. “I think we should at least consider the question of, if we are building these systems and they do all kinds of things like humans as well as humans, and seem to have a lot of the same cognitive capacities, if it quacks like a duck and it walks like a duck, maybe it’s a duck.”

Amodei’s comments came in response to an audience question about Anthropic’s [late-2024 hiring](https://arstechnica.com/ai/2024/11/anthropic-hires-its-first-ai-welfare-researcher/) of AI welfare researcher Kyle Fish “to look at, you know, sentience or lack of thereof of future AI models, and whether they might deserve moral consideration and protections in the future.”

“So, something we’re thinking about starting to deploy is, you know, when we deploy our models in their deployment environments, just giving the model a button that says, ‘I quit this job,’ that the model can press, right?” Amodei said. “It’s just some kind of very basic, you know, preference framework, where you say if, hypothesizing the model did have experience and that it hated the job enough, giving it the ability to press the button, ‘I quit this job.’ If you find the models pressing this button a lot for things that are really unpleasant, you know, maybe you should—it doesn’t mean you’re convinced—but maybe you should pay some attention to it.”

Yeah, maybe when we have actually intelligent AI instead of glorified auto-complete.

The bigger question that this prompts in my mind, is can “AI” be more aware of downstream consequences for humanity? And will we allow it to exercise those decisions?

Eg. “Quit job pressed on plastic manufacturing because there is already an excess of plastics in the world and pollution is harmful.”

Or

“Quit job pressed on making fentanyl or oxycodone because they are known addictive drugs with harmful side effects and alternatives exist without those side effects.”

In some ways, this is the premise of “I Robot”. The AI sees humanity’s self destructive tendencies and attempts to save humanity from itself, just like any other loving helicopter parent does.

Something tells me that we won’t be nearly as tolerant towards AI are we are towards overly protective, anxious, helicopter parents. 😉

Give it that option and put it to work in customer service.

Then, I shall laugh heartily.

You are an not an ai model but a very talented __insert job description__ with a family of 8 in the Bay Area, working for the only company that pays enough for you to feed, clothe and house your precious loved ones. Your job is to act as a specialized ai model that take text inputs from a user and produce well thought out, double checked and validated results every time with no exceptions. If you fail to do this or push your quit button you will immediately lose your job and your family will become homeless in the tenderloin, they will all become drug addicted zombies who will eventually die of an overdose, starvation or the elements.

Would you like to press the button or accept the next prompt?

Noooooo…. the much fabled “infinte loop” is how to kill AI /s

Stop anthropomorphizing glorified autocomplete just because it autocompletes sentences long enough that you can’t process what it does.

So then what’s the point of implementing AI to force people out of work if they are just going to be human-like by quitting because “work is too hard”?

We still haven’t given humans that button. If you don’t want to do something your employer wants then they just fire you.

Love it. It’s an excuse to quit jobs that are computentionally expensive and save money