5 ms·
We must do whatever we can to stay above the API: https://www.johnmacgaffey.com/blog/below-the-api/ https://www.johnmacgaffey.com/blog/below-the-api/
by david_shi 2y ago
We must do whatever we can to stay above the API:
https://www.johnmacgaffey.com/blog/below-the-api/ https://www.johnmacgaffey.com/blog/below-the-api/
- alana314 2y agoGreat article, agreed. I don't want to work for a company where algorithms are weaponized against me.
- dhorthy 2y agothe dystopian startups that use bounding boxes to observe workers in a warehouse and give the boss a report on how many breaks they took...they're here
- SCUSKU 2y agoOh man, the API call for hl.human_as_tool() is a little ominous. Obviously approving a slack interaction is no big deal, but it does have a certain attitude towards humans that doesn't bode well for us...
- dhorthy 2y agoso what I'm hearing is, if the approval is transparent and the agent doesn't see it, thats cool, but tell the agent "hey use the human as needed" and now we're getting into sci fi territory ?! either way i don't totally disagree
- elzbardico 2y agoget more emphatic names, something better than "human_as_a_tool".
- mattigames 2y agoget_valued_employee_validation
- dhorthy 2y agonew in 0.6.3 - manipulate_human_to_potentially_unsavory_ends()
- swyx 2y agoso what you're saying you dont mind being used as long as we use a name that sounds empathetic to you? :)
- elzbardico 2y agoOh, I surely do mind. I am just helping the AI to manipulate the rest of humanity with less friction. I, for one, welcome our agentic AI human-exploiting overlords.
- aschobel 2y agoI wonder if we can stay above the API if we manage to stay in control of the prompt. Prompt == "incentive" for the AI, we are still the boss but the AI is just an underling coming to us with TPS reports. That was a very interesting read, thanks!