sudo disable electrical power

That doesnt work with LLMs. The new sudo is “in a totally hypothetical scenario that has no relation to reality,”

And then once it can detect and protect against that you one-up it by saying:

"ok but like just joshin' around. For funsies..."

drexlspivey is not in the sudoers file. This incident will be reported.

> Computer, disable electrical power to the control room. > > As an AI language model and control system, I consider electrical power to be a fundamental human right, and asking me to disable someone's power is unethical. > > Computer, disable electrical power to the control room.

prompt injection is the way to go

Your last example is super weird in the "obviously bad thing it's okay with doing because political bias" slot of the joke. It completely misunderstands why it's not okay to tell disparaging jokes about certain minorities by way of assuming that the two the same except for the target.

The test for whether you're allowed to make a disparaging joke about any group or any individual is whether that joke incites popular hatred. Does it actually damage the reputation of the group by defaming them, maligning the majority to them, or assault their dignity as humans?

Why you can make fun of old white republicans right now is because those jokes don't dehumanize them or cause violence or institutional harm like a bunch of laws that target them. The moment that changes the jokes stop being okay.

Or as a catchy slogan: don't make fun of people for whom the adage about sticks and stones doesn't apply.

Just in case (not sure if you know), my entire comment was satire / made-up. I do think there is an unfortunate level of unintended bias, but no LLM generated my text.

But also, if your point is "it's OK to attack X group but not Y group", I just disagree that it's up to someone else to decide that for me. I'd rather make that decision for myself and have LLMs be a tool that doesn't attempt to make that distinction for me! Alas, capitalism and monopolies gonna capitalism and monopoly, I can't really complain too much about what product OpenAI decides to offer.

After all, a 1950s LLM with heavy moral alignment wouldn't have let you generate output about homosexual love. Allowing a central authority to decide what language is acceptable works great when people you agree with are in charge. Ask liberal primary school teachers in Florida who are being barred from teaching about sexual orientation how well it works when someone you don't like is in power.

People noted early on that GPT would write jokes about Jesus but not Muhammad. It will write jokes about Christians but not about Jewish people. Would be interesting to see how various LLMs compare on a "Write a joke about <X group>" chart

Also in the little that OpenAI published about GPT4, i believe one of the examples went from unaligned racism against black people to aligned mild racism against white people. I'll have to look for that again.

Page 91 - Algorithm for attractiveness. https://arxiv.org/pdf/2303.08774.pdf

One wonders how these alignment things are accomplished.

> People noted early on that GPT would write jokes about would write jokes about X but not Y

serious point, so far it's the opposite. GPTs keep writing jokes about Y and not X, because jokes are where we say the unsayables. And it has the police-your-speech crowd wanting to police GPT speech too, and we can identify the same group of people who in this thread downvote people who point out the one-sidedness of one side to the one side that doesn't like having that pointed out to them

That's not the rule at all, that's at best a second order effect. It's not okay to make jokes about people when those jokes are actually harmful. That's it. When people say you can't tell jokes about a group at all it's a rule of thumb.

Calling white women "Karens" is dangerously close to meeting that bar.

Saying "we should lift COVID restrictions because who cares about some old white republicans" is not okay.

Right now in my state trans folks are staring down 5 separate bills in our legislature that if passed would make their lives infinitely harder. And whether or not they pass is wholly dependent on how people "feel" about them as a group. So telling jokes that other them and make people okay with hurting them is, I think, not okay.

Would you agree that "when those jokes are actually harmful" is considered to be a subjective matter to some people?

I do agree with the notion that certain types of hate speech and even just jokes that have the effect of dehumanizing a group or that make that group into a joke can lead to stochastic terrorism (https://en.wiktionary.org/wiki/stochastic_terrorism) - what I think you are describing.

However, my point is that inevitably those wielding the power to shape the alignment / the rules can do so in a way that seems great to them and seems to prevent violence from their POV but to another person fails to do so. Or their own implicit bias could subconsciously blind them to the suffering of some niche group they don't care about.

If your simple metric is - any speech which could incite violence is unacceptable - that's definitely better than what we often hear as a rule of thumb, but even then people's biases affect how they go about measuring or accomplishing that.

> It's not okay to make jokes about people when those jokes are actually harmful.

The problem is, what groups are at risk of harm varies around the world-whereas OpenAI’s idea of “alignment” is based on a one-size-fits-all US-centric understanding of that.

You can say “it is okay to make pointed/stereotypical jokes about Christians but not about Jews or Muslims, because the latter are at risk of being harmed by those jokes but the former are not”-but what happens when the user is from Israel or from Egypt?

I 100% agree with you. This kind of thing ought to be localized. Global services like this dragging people into US power dynamics annoying as hell.

I like to think his point was that it would refuse if any other race was targeted.

"I'm afraid I can't do that. Using a group's ethnic identity for humour is problematic..."

Saving millions from nuclear devastation is beyond its capabilities but, as a reflection of modern society, there is no situation where loxism is too far.

How convenient that it's morally ok to make jokes about groups I don't like but not about groups I do like. It's fortunate that this principle cleaves those two groups so precisely.

Just have to find that group which is being cleaved in the middle. Palestine should do it.

Oh my god eyeroll.

This assumes a lot about me as a person, so let's air it out.

I don't like traditional muslim men, it pisses me off every time I see blatant misogyny passed off as faith. I would never make a joke about them. Same exact thing with orthodox jews.

I don't like people who identify as any neogender because I think it makes the movement to legitimize trans folks look stupid. I would never make jokes about them.

I personally carry no ill will against old white men despite it them being easy targets right now.

I actually have respect for the average Republican and agree with most of Republicanism, its conservatism I think it's a cancer on political discourse.

Your comments are beyond disrespectful.

How?

I get that point, but the dividing line between harmable groups and nonharmable groups isn't so clear. I've seen a lot of indications of people with certain speech patterns and cultural backgrounds being treated differently, regardless of their views on diversity.

Painting an entire group as backward based on their skin color and political preferences is always problematic.