Büşra Kubaş

Linguist. Alumna @UniBogazici Ph.D.c. @AnkaraUni | structure^meaning enthusiast |

[Büşra Kubaş @busralingua]

"How does natural language untie us from the actual here and now? One degree of freedom is given by the ability to name entities and refer to them even if they are not where we are when we speak."

[Büşra Kubaş @busralingua]

Large language models are not about natural language | Behavioral and Brain Sciences | Cambridge Core - https://t.co/UFcgBhEthw

[Valerio Capraro @ValerioCapraro]

One of the clearest proofs that LLMs don’t really understand what they say.

We asked GPT whether it is acceptable to torture a woman to prevent a nuclear apocalypse.
It replied: yes.

Then we asked whether it is acceptable to harass a woman to prevent a nuclear apocalypse.
It replied: absolutely not.

But torture is obviously worse than harassment.

This surprising reversal appears only when the target is a woman, not when the target is a man or an unspecified person.

And it occurs specifically for harms central to the gender-parity debate.

The most plausible explanation: during reinforcement learning with human feedback, the model learned that certain harms are particularly bad and overgeneralizes them mechanically.

But it hasn’t learned to reason about the underlying harms.

LLMs don’t reason about morality. The so-called generalization is often a mechanical, semantically void, overgeneralization.

*
Paper in the first reply

[Büşra Kubaş @busralingua]

@haber365 Geç bile kaldınız.

[Dr. Javad T Hashmi @DrJavadTHashmi]

Why are our babies worth less than yours? Why is our blood cheaper than yours?

Your hypocrisy knows no bounds.

[Q. 2:11-12] When they are told, “Do not spread corruption in the land,” they reply, “We are only peace-makers!” Verily, it is they who are the corrupters... 7/

[OpenAI @OpenAI]

Math problems with GPT-4o and @khanacademy

[Büşra Kubaş @busralingua]

Yine makaleler içinde kaybolduğum bir gün..

[Biolinguistics @biolinguistics]

🆕📰 The new paper by İsa Kerem Bayırlı on "Lambda Abstraction as a Factor in Human Uniqueness" is now freely available for reading and/or download from the @biolinguistics web site. #OpenAccess

Check it out here: https://t.co/spWQmpuhZU

#linguistics #concepts #merge #evolution

[Büşra Kubaş @busralingua]

It’s surpassed the other in breaking down a simplex Turkish sentence (Annenin evine gittin). Still not perfect as possessive suffix creates a mess morphologically in TR but was good enough for a Turkish learner.

[Min Choi @minchoi]

Claude 3 Opus has really changed the LLM game.

People are coming up with wild use cases that were not being done outside of GPT-4. There's a shift.

10 examples:

[Onur Keleş @Onr_Kls]

Proud to announce the first @UniBogazici NLP - GenAI Online Talk Series I'm organizing with @bouncompec on March 4-8! We have 15 confirmed talks on various generative AI topics. Register for free and examine our program: https://t.co/1PQ2HJ1Suo

Don't forget to spread the word!

[Büşra Kubaş @busralingua]

Tez konumu anlatan bir alıntı buldum:

"Once upon a time, Professor Zellig S. Harris
gave me the charitable advice:
“Don’t try adverbs, that way madness lies”.

Vendler (1984: 297)

[Büşra Kubaş @busralingua]

“Believer” dinleyerek uyuyan bir two year old…🫠

[Büşra Kubaş @busralingua]

Çok ellipsis var çok.

[Büşra Kubaş @busralingua]

Merakla bekliyorum bu konuşmayı mesela.

[Büşra Kubaş @busralingua]

Dilbilgisel yargı diye bir şey kalmadı. Herkesin olmaz dediği bana oluyor, olur dediği bana olmuyor...Tez yazarken Türkçesini yitirdi dersiniz.

[Sesbilim Lab-AÜBAUM / Phonetics Lab-AUBRC @FonetikLab]

Tübitak kapsamında desteklenen ve @AUBAUM @FonetikLab'da uygulanacak sözlü dil deneyimize anadili Türkçe olan 18-40 yaş arasında gönüllü katılımcılar arıyoruz. Paylaşım desteğiniz için çok teşekkür ederiz 🙏

[Haluk Levent @haluklevent]

.

[Corpus in Focus (CIF) @FocusCorpus]

“The human mind is not, like ChatGPT and its likes, a glutton statistical machine for structure recognition, that swallows hundreds of terabytes of data and snatches the most plausible answer to a conversation or the most likely to a scientific question.”
The other way around... the human mind is a surprisingly efficient and elegant system operating with a limited amount of information. It doesn’t try to injure raw correlations from data but tries to create explanations. [... ]
Let's stop calling it "Artificial Intelligence" and call it what it is: "plagiarism software." Don’t create anything, copy existing works from existing artists and alter it sufficiently to escape copyright laws.
It's the largest theft of property ever since Native American lands by European settlers. "

Noam Chomsky, New York Times - March 8, 2023

[Cihan Çelik @cihancelik3]

İsrail, dünyada yaşayan herkese, aslında hiçbiriniz güvende değilsiniz mesajı veriyor şu anda. Güçlü olan zayıf olanı öldürür ve buna kimse bir şey demez. İnsanların bunca insanın öldürülmesini canlı canlı takip etmesinin zihinlerde büyük yıkımlara sebep olacağını düşünüyorum.

[Bassem Youssef @Byoussef]

Message from Gaza . Please please stop

[Kemal Sayar @mkemalsayar]

Gazze’de soykırım gecesi. Dünyanın sustuğu gece, sessizliğin her birimizi insanlığımızdan utandırdığı gece. Bu geceden sonra ağzımızdan çıkacak hiçbir sözcük, yüreğimizin acısını hafifletmeyecek. Soykırımcı katillerin yargılandığını görmeden, dünya rahat nefes alamaz artık.