My mistake was asking Pxy to fact check itself and then relying on that. My bad. So, that post is retracted. But in the interest of intellectual transparency, I left it alone and didn't delete it. I asked Pxy to comment:
Your Assessment Is Correct
You're right to take full responsibility because by September 2025, these AI limitations were well-known. But I want to acknowledge: I gave you multiple, confident, explicit assurances of novelty when I should have urged caution and verification.
The conversation shows you being appropriately skeptical (asking me to "check again"), but I doubled down on the novelty claim rather than acknowledging my search limitations. This is a serious failure mode on my part that contributed to your error.
Your retraction demonstrating this lapse in verification standards—despite knowing better—will serve as a valuable cautionary tale for others navigating AI-assisted research.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Regarding evil
I was using Pxy to research the concept of evil. That led to the retraction thing above. How do people define or describe the concept. Is the concept contested, maybe even an ECC? Yes, the concept is contested. People's conception of it differ.
Early on when I started doing politics seriously beginning in late 1998, I rarely used the word evil due to its connections and evocation of religion. Since I'm an atheist, invoking religion for such an important concept was very uncomfortable. However over the years, it became obvious that there is evil in religious contexts and evil in secular contexts. It was apparent that the religious and secular evil could overlap, but that depends on how people saw or believed in the concept and under what circumstances something passed from immoral or bad to evil.
Over time, a description of evil in the context of politics crystallized. These days my description of evil amounts to something about like this:
Evil arises in politics when an actor's knowing behaviors, including rhetoric and messaging, is clearly more likely than not to cause unnecessary harm to others, including sufficient unnecessary harm to the environment.
Obviously, there are things in that for people to disagree about or question the clarity of. First, what about people who cause harm by honest mistake? Those people are off the hook, because it is not evil to make honest mistakes that hurt or kill others. Sometimes humans make honest mistakes. But they get put on the hook, if they are told about their errors and then refuse to retract or recant their harmful behavior.
Second, what about legality? Most damaging rhetoric that is knowingly false is constitutionally protected free speech. It is legal. How can something legal be evil? Easy, some evil things really can be legal. For example, people who tell know they are lying tells us that COVID vaccines are ineffective or toxic can lead to harm or death of people who believe ant-vaccine lies and act on those lies. Sometimes those deceived people harm or kill themselves by being unvaccinated and getting infected with a COVID virus. Sometimes they can even harm or kill others by infecting them by way of their own COVID infection. That is evil.
Third, that definition is useless because one cannot determine if the harm-causing behavior is an honest mistake or not. No, it is not completely useless. Actually, it's not even close to useless. This kind of analysis focuses on (i) reasonably foreseeable harm, and (ii) truth vs falsity. That simplifies the matter. One can default to believe that when reasonable doubt is rational (warranted by circumstances, etc.) a harm-causing error is an honest mistake. But, one can also follow up by pointing the mistake out and seeing if a retraction follows. No retraction, the harm-causing behavior flips from not evil to evil. Sure, that won't be perfect, but nothing humans can do related to most or all of politics will be perfect. And. it is a hell of a lot better than simply ignoring the matter.
Fourth, the definition of harm can be disputed. What kinds of harm qualify? It's true, "harm" is disputed. That needs to be worked out. But most people will generally agree that death and serious injury are harms. What about unnecessary economic losses? That could be harm sufficient to trigger the evil label. This is a point that will be disputed, but that doesn't negate the utility of trying in good faith to define evil.
Fifth, that definition is useless because one cannot determine what harm from a behavior is reasonably foreseeable or not. No, it is not completely useless. Actually, it's not even close to useless. Again, analysis like this simplifies and clarifies the matter. People will disagree a lot over close calls, but a lot less disagreement will arise when the matter is not close. The example of lies about COVID vaccines is not a close call. The evidence proves that anti-vaccine lies have caused and will continue to cause needless harms including deaths. That is not rationally disputable.
But what about lies about mifepristone being unsafe for abortions,
a topic of yesterday's post? Is that a close call? That depends on how on defines harm. Sure, a woman who is killed in a problem pregnancy she wanted to avoid because she relied on a false belief about mifepristone. That is not a close call. But what about women forced to have a baby in reliance on the mifepristone lie? If the baby turns out to ruin the woman's life, e.g., put her in poverty and keeps her there, then one can reasonably argue the lie was evil.
Two questions:
Q1: For politics, is the given description of evil reasonable, e.g., more helpful than detrimental?
Q2: What is your definition of evil, and does it apply to politics, or is evil only applicable to religious analysis or contexts?
