Mastodawn

Claus Aranha 5d ago

Today I learned that, apparently, "Zero Shot" just means "we separate train and test data".

I hate ML

Claus Aranha 6d ago

Professor "I" is visiting me in a few weeks to discuss student exchange.

"I": By the way, could you introduce me to professor "J" during the visit? He does the same kind of research that I do.

Me: Of course! Can you give me a link to your webpage for me to put it on the introduction e-mail?

"I": <sends a google scholar link>

Me: <screams internally>

People, there is NO EXCUSE for a CS prof/researcher to not have a webpage. Gah.

Claus Aranha Jan 22

The person making the IEEE template for conferences is tired of your shit:

```
{\BibTeX} does not work by magic. (...)

{\LaTeX} can't read your mind. (...)

{\LaTeX} does not have precognitive abilities. (...)
```

Claus Aranha Jan 22

This week I've been invited to review for two conferences that didn't bother saying when the review period would be.

Please don't do that, folks.

Claus Aranha Jan 15

`Xadrez Verbal` is a podcast hosted by two historians that makes weekly round-ups of international news.

Today they released a 4-hour chat on the Venezuela-US relations, done right: Going all the way back to the Venezuelan independence in the early 1800s, and reviewing how it the events over its history reflects on what is happening today.

You can't understand the present if you don't know the past.

https://www.youtube.com/watch?v=A0h-vInVex0

(also highly recommend their weekly program)

TUDO sobre a crise na Venezuela e as relações com os EUA - XADREZ VERBAL

YouTube

Show thread

Claus Aranha Jan 8

PS 1: This is about AI, but this is not only about AI.

PS 2: Yes, one can think of situations where incentives to lie or badly formed rules can make communities worse. Not the point of the thread.

Show thread

Claus Aranha Jan 8

One final thing that rules and CoC can do is to encourage people who agree with the code, but feel isolated, to feel like they are welcome to contribute their ideas.

In one of the conversations I was part of, one person responded to the suggestion of explicit anti-AI rules with something along the lines "I have been thinking something similar, but didn't say anything because I thought I was the only one"

Another reason for "unenforceable" rules. (end)

Show thread

Claus Aranha Jan 8

Even if one lies about following a rule, and gets away with it, I'd say that there is still some cost for the successful liar.

In an environment where several communities co-exist, each with their own rules, there is an attrition to continuously lying about what you're doing, and an incentive to joining a community where the rules better match your way of thinking.

In this way, behavior rules (no AI code contributions), help shape community. (3/?)

Show thread

Claus Aranha Jan 8

There will always be someone who will lie about following a rule or law. That is not a good reason for not having the law in the first place.

At the very least, when you do catch someone lying, you have the apparatus to punish / banish / shame the perpetrator.

But, more importantly (IMO), a rule/law establishes the character of the community. "This is what we are", "This is what we think is right", "This is what we aim to be".

(2/?)

Claus Aranha Jan 8

Recently I've been part to different conversations about how to reject AI in several separate communities.

The discussions were often similar: Someone proposed banning AI on [writing/reviewing/coding]; someone else replies: "that won't work, people will lie about it and we won't be able to detect / it will be too much work to detect".

I think that's a bogus argument. Rules exist not only for enforcement, but also for community building. (1/?)

Work Site	http://conclave.cs.tsukuba.ac.jp/
Languages	en, ja, pt, fr
Pronouns	They