Here's the incident page
https://status.openai.com/incidents/ssg8fh7sfyz3
2,511 karma · joined June 25, 2015
> "Yeah you won’t figure those APIs out from the documentation. It was on purpose. You have to go buy the book."
"Invites are currently paused for this server. Please try again later."
I currently feel like it's too out there to comfortably propose to companies I'm interviewing at. If a job description explicitly stated "open to discussing part-time opportunities" and leadership was realistically aligned with the time commitments part-time engineers would face, I would definitely bring it up as a possibility.
There's so many companies doing cool work. Individually we commonly end up contracting our aspirations in service of one endeavor. That makes sense for some companies. For others, the benefits in terms of production value and having their engineers exposed to multivariate problems could be huge. I'm particularly thinking if companies that have straightforward apps to present or deliver their services, like a video app, community apps, etc. A lot of those problems are solved, and I'm not sure it makes sense to invest a huge amount, in terms of full-time work, in executing on those visions.
I'm also curious what it would be like to manage a part-time engineer. Work life balance would have to be front and center. Processes for dealing with emergencies would need to be realistic. Demands from management would need to be broken down with part-time chunks in mind, and so on.
I base my understanding of generating internal motivation of of the WOOP model, academically known as implementation intentions with mental contrasting. Focusing on mounting my internal dialogue on simple if-then models is easier than running through the whole WOOP model.
I'd love to know of any mental heuristics others here use that help them effectively close the current state vs ideal state gap.
Since you expressed openness to this dialogue, I'll share. When I really, really stretch my moral imagination to the point it decouples from practical logic, I find myself feeling spiritually reverent for the things that the mental abstraction "complexity" points at. I arrive at this by synthesizing a particular metaphor from Alan Watts with Mihalyi Czikszentmihalyi's ideas on complexity.
In the same way an apple tree apples, the universe peoples. It also dogs, and weathers and softwares and cows. Of it emerge these beautiful structures, including plants. If I was capable of surviving the endeavor, cherishing and protecting every single one of those things that the universe expresses would be my ultimate ethical goal. I don't know if there's farther expansive rings of morality. I imagine I might be able to peek at some with the help of deep, deep meditative practice or other tools.
The thing that re-embodies me is the Buddhist idea of bodhisattvas. They are beings who have reached enlightenment, but then rejected paradise in honor of the work still left to do on this plane of existence.
That's basically the outline of my metaphysical beliefs.
Finally, specifically regarding pain and plants, I agree that we bend our intuitions to the experiences that make sense to us. My argument is that the word "pain" and what we understand pain to be, is predicated on a specific biological structure, nociceptors. To speak of pain without nociception is like saying the water in my hand isn't wet. They're enmeshed, dependent properties. So, if plants do experience something that increments the universe's suffering counter, which I leave as a possibility, we would have to imagine some other form of communication. Language and thought extrudes from the human body. As a consequence of that fact, there are just some things it can't capture, and attempting to do so places us closer to the choice: bodhisattva or buddha.
Quantum computers have special properties that make them capable of breaking commonly used encryption schemes. We're dependent on those schemes for secure communication, like logging into a bank website. Due to this, the organization NIST has been working on finding encryption schemes that would be diffcult to break with a quantum computer.
NIST was at the stage of this project where they felt reasonably confident that three specific encryption schemes met the requirements. This is after review of many candidates. Rainbow made it to the top three. Ward Buellens essentially managed to break Rainbow, thereby making it ineligible for use in a quantum computer-powered future.
I assume this puts the two remaining candidates' eligibility into question. Were the requirements lacking, or is the project inherently at risk of failure due to the nature of quantum computing?
My basic formula for moral truth is: starting from a non-maladaptive state of mind, would I prefer to have or recoil from 'X' experience? With that in mind, could I reasonably conclude that sentient others around me would arrive at the same conclusion as me? How about strangers? Other animals? An extraterrestrial? A synthetic being? Humans in a million years?
If my chain of reasoning doesn't break at any point as I expand the set of sentient beings to check against, then it is likely that I now have clear guidance on engaging in actions that produce experience 'X'.
Here is an example: Should I break my friend's arm against his will? I would not want that to happen to me. Most people I know prefer not to have their arms broken. My dog would not want that. A cow would not want that. A fish would not want his tail snapped off. People in the future won't want their arms broken if it feels how I imagine it feels. I don't imagine that a sentient robot would like its gripping mechanism broken. Whatever sorts of appendages extraterrestrials have, they probably don't want to lose the ability to use them.
I can then extract the generalizable term from "I shouldn't break my friend's arm against his will" to, "I shouldn't hurt sentient beings when it's counter to their overriding preferences."
It now follows that if I were able to poll every sentient creature through all time (sentient according to a rubric), I would expect the majority to share the uncovered preference or aversion in question. I now have a positive statement about morality that is universalizable, apparently akin to Kantian philosophy, which I have some bare surface knowledge of.
The exercise seems pretty robust to me insofar as language and idea-derived conclusions hold up. For example, culturally dependent preferences, idiosyncrasies, aesthetic preferences, etc, fail the test easily.
I won't bury the lede further. Plants can't experience pain, at least insofar as we conceive of it, because they don't have nociceptors that signal pain. Therefore, if they do have any sort of phenomenological experience, whether it would be recognizable to us or not, there is no sense in describing anything they experience as "pain".
There may be simple damage-avoidance reflexes, but comparing that with what it feels like to break a bone would be like saying that pedestrians and vehicles should share the road. The difference in quality is so significant that it makes better sense to categorize those reflexes as not pain at all although in service of the same tissue-protecting goal. My legs are not a vehicle I don't think.
A final point on the deficit of nociception, is that the experience of pain is so central to our lived experience that it's difficult to conceive of what that would be like. It's like the alpha channel in rgba... Without that signal for opacity, there's no color at all.
If the above is not enough to convince, there's some recent research that sought to systematize the analysis of sentience. Here's the 8 points used:
1) possession of nociceptors;
2) possession of integrative brain regions;
3) connections between nociceptors and integrative brain regions;
4) responses affected by potential local anaesthetics or analgesics;
5) motivational trade-offs that show a balancing of threat against opportunity for reward;
6) flexible self-protective behaviours in response to injury and threat;
7) associative learning that goes beyond habituation and sensitisation;
8) behaviour that shows the animal values local anaesthetics or analgesics when injured
I hope that helps!
An interesting aspect of the internet is that it makes the determinant lines fuzzier and clearly arbitrary. Objectively, why would the life of a child in one country matter more or less than one from the other side of the planet? It's completely arbitrary, but the entrenched human social structures have been built around physical proximity and so we end up donating money to feed a relatively healthy child instead of donating to the distant one on the verge of dying from malnutrition.