Lightnews — Scholar-powered news

Mel Andrews

@bayesianboy.bsky.social

11K followers 4.6K following 710 posts

I’m not like the other Bayesians. I’m different. Thinks about philosophy of science, AI ethics, machine learning, models, & metascience. postdoc @ Princeton.

Posts Media Videos Starter Packs

Pinned

Mel Andrews @bayesianboy.bsky.social · Nov 17

A thread of my papers. The first was recently published with Cell Press, coauthored with Andy Smart and Abeba Birhane. It tackles the connections between epistemic deficiencies in deployed ML and ethical consequences. www.sciencedirect.com/science/arti...

The reanimation of pseudoscience in machine learning and its ethical repercussions

The present perspective outlines how epistemically baseless and ethically pernicious paradigms are recycled back into the scientific literature via ma…

www.sciencedirect.com

9 23 150

Reposted by Mel Andrews

Shahab Bakhtiari @shahabbakht.bsky.social · 12h

Regardless of what explainability/mech interp in AI is actually after, and whether or not they know what they’re searching for, we can confidently say they’re pursuing what systems neuroscience has pursued for decades, with very similar puzzles and confusions.

Mel Andrews @bayesianboy.bsky.social · 13h

What problem is explainability/interpretability research trying to solve in ML, and do you have a favorite paper articulating what that problem is?

1 8 42

Mel Andrews @bayesianboy.bsky.social · 9h

Something in the realm of a functionalist approach to explanation in the domain of explainable ML was exactly what I was seeking after. Thank you (for sharing it and for writing it!).

Mel Andrews @bayesianboy.bsky.social · 12h

Oh, this is fascinating. I’ve attempted to carve out intellectual/sociological/historical divisions in this space a number of ways, and that one has not been apparent to me! I would love to pick your brain on this.

1 1

Mel Andrews @bayesianboy.bsky.social · 13h

Excellent. Thank you!

Mel Andrews @bayesianboy.bsky.social · 13h

What problem is explainability/interpretability research trying to solve in ML, and do you have a favorite paper articulating what that problem is?

11 5 28

Mel Andrews @bayesianboy.bsky.social · 1d

anyone invented autoblock for this site (feature that allows you to block yourself from seeing anything that you yourself have posted)?

2 13

Reposted by Mel Andrews

anti-blinding-headlight advocate @barneyretina.bsky.social · Jan 23

A user of the anti-blinding-headlight advocacy group at reddit.com/r/fuckyourheadlights tested headlight brightness on multiple vehicles in order to figure out what's going on with regulations and adherence to them.

Here's a 4-part infographic with key findings:

1 11 24

Mel Andrews @bayesianboy.bsky.social · 1d

What is then the essence of the controversy?

Mel Andrews @bayesianboy.bsky.social · 1d

it to your moral character. I think it is intrinsically deceptive and therefore immoral to ship a product that seems human-like, and that it is actively bad for you to treat these products like they are humans.

1 3

Mel Andrews @bayesianboy.bsky.social · 1d

I do believe you that that dynamic is happening and it sounds godawful and I am thankful to have avoided exposure to it thus far. I still do not think that there is anything special about modern software that it deserves special pseudo-moral status or that we can draw inferences from how you treat

3 2

Mel Andrews @bayesianboy.bsky.social · 1d

It is possible that people are using derogatory language aimed at computers to behave racistly (???) which means that people can be racist but still does not mean that a bad word about a computer is a slur because that isn’t what the word “slur” means.

1 11

Mel Andrews @bayesianboy.bsky.social · 1d

Right. And that’s bad. But using a bad word about a computer is actually literally not a slur.

Mel Andrews @bayesianboy.bsky.social · 1d

I also had not heard that term before. Clearly behind on the discourse.

To Jeremy, yes, that and also that it’s weird to pretend that we can’t use derogatory language about anything at all, ever, without descending into something that is equivalent to actual racism?

1 1

Mel Andrews @bayesianboy.bsky.social · 1d

I am hoping that this isn’t what Neil meant and choosing to interpret charitably: for something to be a slur it has to actually refer to a group of people and I don’t think we have evidence that people use that term to refer derogatorily towards groups of people.

2 1

Mel Andrews @bayesianboy.bsky.social · 1d

Certainly. But I think it is very important here to keep these distinct and keep in view what an actual slur is.

Mel Andrews @bayesianboy.bsky.social · 1d

I broadly agree.

Mel Andrews @bayesianboy.bsky.social · 1d

1. I don’t think it’s a bad habit to be mean to objects.
2. I think people are good at finding ways to be mean to people and use objects in these endeavors in weird ways.
3. Anthropomorphizing an object and being mean to it just isn’t the same as violence towards a human being.

2 2

Mel Andrews @bayesianboy.bsky.social · 1d

hang on I missed this and I have to know about this discourse

Mel Andrews @bayesianboy.bsky.social · 1d

If we see a person behaving towards a sex doll in some particular way we might draw inferences about his treatment of women. We might even suspect him of being a misogynist. We would not therefore say that he had assaulted, violated, or even mistreated the inanimate object. Because it’s not a person

1 1

Mel Andrews @bayesianboy.bsky.social · 1d

1 2

Reposted by Mel Andrews

M R. X. Dentith @mrxdentith.com · 1d

A frustrating thread (I agree with @bayesianboy.bsky.social here) in which Mel tries to stick to the point that slurs harm people not objects and people keep saying "But people sometimes use the same word but in different contexts, so therefore you are wrong..."

Mel Andrews @bayesianboy.bsky.social · 1d

Let me make this really simple for those of you who don’t understand: a slur is a term that is harmful because its referent is a subject who can be harmed by it.

SE Gyges @segyges.bsky.social · 1d

i wonder which of these two posts on her pro slur thread timnit blocked me for

2 1 16

Mel Andrews @bayesianboy.bsky.social · 1d

Right. I think that this is deeply cringe, but not necessarily racist or motivated by a desire to act in a racist manner.

Mel Andrews @bayesianboy.bsky.social · 1d

It is very weird to me to pretend that all derogatory language use towards anything is some kind of morally slippery slope because some kids online are making jokes in poor taste or whatever it is they are doing, which is very obviously a shift from the linguistic reference under discussion

3 4

Mel Andrews @bayesianboy.bsky.social · 1d

it is also, I repeat, definitionally not a slur, because a slur is a notion with an agreed upon meaning

Mel Andrews @bayesianboy.bsky.social · 1d

But still not constitutionally what a slur is. Because the referent is not an identity-sharing group of people.