#Tag · bonfire.cafe

Nick Byrd, Ph.D.

@ByrdNick@nerdculture.de · 7 days ago

How well can #AI infer the relationships between a pair of answers?

Are both plausible? (this AND that)
Is one more plausible? (this OR that)
Are both implausible? (NEITHER this NOR that)

LogicalCommonSenseQA benchmarks #LLMs on such #logic inference.

https://doi.org/10.48550/arXiv.2601.16504

"(1) all models perform reasonably well on AND and moderately on OR, and (2) performance collapses on NEITHER/NOR in zero-and few-shot settings, even for the strongest LLMs. ...weaknesses are amplified in the MIXED condition, where different operators appear across options and models must implicitly infer the governing relation. F1 drops to 43–56% for all few-shot models, [but] fine-tuned models achieve 83–93% F1 across all operators, suggesting that the task is learnable with supervision...."

"While LLaMA-3.1-8B achieves 72.2% accuracy on CommonSenseQA, its performance drops sharply on Logical-CommonSenseQA: 72% on AND, 62.2% on OR, 42.7% on MIXED, and only 13.9% on NEITHER/NOR. This discrepancy shows that benchmarks like CommonSenseQA substantially overestimate models’ commonsense reasoning ability by failing to test relational and compositional plausibility judgments." — "(1) all models perform reasonably well on AND and moderately on OR, and (2) performance collapses on NEITHER/NOR in zero-and few-shot settings, even for the strongest LLMs. ...weaknesses are amplified in the MIXED condition, where different operators appear across options and models must implicitly infer the governing relation. F1 drops to 43–56% for all few-shot models, [but] fine-tuned models achieve 83–93% F1 across all operators, suggesting that the task is learnable with supervision...." "While LLaMA-3.1-8B achieves 72.2% accuracy on CommonSenseQA, its performance drops sharply on Logical-CommonSenseQA: 72% on AND, 62.2% on OR, 42.7% on MIXED, and only 13.9% on NEITHER/NOR. This discrepancy shows that benchmarks like CommonSenseQA substantially overestimate models’ commonsense reasoning ability by failing to test relational and compositional plausibility judgments."

arXiv.org

LOGICAL-COMMONSENSEQA: A Benchmark for Logical Commonsense Reasoning

Commonsense reasoning often involves evaluating multiple plausible interpretations rather than selecting a single atomic answer, yet most benchmarks rely on single-label evaluation, obscuring whether statements are jointly plausible, mutually exclusive, or jointly implausible. We introduce LOGICAL-COMMONSENSEQA, a benchmark that re-frames commonsense reasoning as logical composition over pairs of atomic statements using plausibility-level operators (AND, OR, NEITHER/NOR). Evaluating instruction-tuned, reasoning-specialized, and fine-tuned models under zero-shot, few-shot, and chain-of-thought prompting, we find that while models perform reasonably on conjunctive and moderately on disjunctive reasoning, performance degrades sharply on negation-based questions. LOGICAL-COMMONSENSEQA exposes fundamental reasoning limitations and provides a controlled framework for advancing compositional commonsense reasoning.

Ulrike Hahn and 1 other boosted

awhamburg

@awhamburg@wisskomm.social · last week

🎙️ Our Young Academy Fellow Dr Deniz Sarikaya launches ‘aboutlogic’
+++
In the new biweekly #podcast #logic, #mathematics, #philosophy and #computer #science come together, funded by the Academy of Sciences and Humanities in Hamburg.

Watch #videos
👉 https://www.youtube.com/@aboutlogic/videos

Listen to the #Podcast
👉 https://aboutlogic.podigee.io/

@DenizPhiMa

3 media

YouTube

aboutlogic

aboutlogic features bi-weekly conversations on all topics around logic: From foundations of computer science, set theory, critical thinking, riddles to the questions how it is to be an academic researcher. Hosts: Thorsten Altenkirch, Deniz Sarikaya Creative Production: Jan-Niklas Meyer Upcoming guests (TBC): Graham Priest — January 28, 2026 Steve Awodey — February 11, 2026 Colin Rittberg — February 25, 2026 Alexander Steen — March 11, 2026 Jordi Fairhurst & Seunghyun Song — March 25, 2026 José Antonio Pérez-Escobar — April 8, 2026 Deborah Kant — April 22, 2026 Urs Schreiber — May 6, 2026

awhamburg

@awhamburg@wisskomm.social · last week

Watch #videos
👉 https://www.youtube.com/@aboutlogic/videos

Listen to the #Podcast
👉 https://aboutlogic.podigee.io/

@DenizPhiMa

3 media

YouTube

aboutlogic

Nick Byrd, Ph.D.

@ByrdNick@nerdculture.de · 2 weeks ago

If your objection to something is just “it's biased”, then you do not yet have an objection.

I mention #theBiasFallacy in "Strategic Reflectivism":
https://doi.org/10.48550/arXiv.2505.22987

and explain it on my #blog: https://byrdnick.com/archives/11072/the-bias-fallacy

#logic #argumentation #debate #edu #criticalThinking

4 media

"'They’re biased, so they’re wrong!' That’s a fallacy. We can call it the bias fallacy. Here’s why it’s a fallacy: being biased doesn’t entail being wrong. So when someone jumps from the observation that So-and-so is biased to the conclusion that So-and-so is wrong, they commit the bias fallacy. It’s that simple.

In this post, I’ll give some examples of the fallacy, explain the fallacy, and then suggest how we should respond to the bias fallacy."

https://byrdnick.com/archives/11072/the-bias-fallacy — "'They’re biased, so they’re wrong!' That’s a fallacy. We can call it the bias fallacy. Here’s why it’s a fallacy: being biased doesn’t entail being wrong. So when someone jumps from the observation that So-and-so is biased to the conclusion that So-and-so is wrong, they commit the bias fallacy. It’s that simple. In this post, I’ll give some examples of the fallacy, explain the fallacy, and then suggest how we should respond to the bias fallacy." https://byrdnick.com/archives/11072/the-bias-fallacy

Nick Byrd, Ph.D.

The Bias Fallacy

“They’re biased, so they’re wrong!” That’s a fallacy. Why? Because being biased doesn’t entail being wrong. In this post, I’ll explain the bias fallacy, gi…

arXiv.org

Strategic Reflectivism In Intelligent Systems

By late 20th century, the rationality wars had launched debates about the nature and norms of intuitive and reflective thinking. Those debates drew from mid-20th century ideas such as bounded rationality, which challenged more idealized notions of rationality observed since the 19th century. Now that 21st century cognitive scientists are applying the resulting dual pro-cess theories to artificial intelligence, it is time to dust off some lessons from this history. So this paper synthesizes old ideas with recent results from experiments on humans and machines. The result is Strategic Reflec-tivism, the position that one key to intelligent systems (human or artificial) is pragmatic switching between intuitive and reflective inference to opti-mally fulfill competing goals. Strategic Reflectivism builds on American Pragmatism, transcends superficial indicators of reflective thinking such as model size or chains of thought, applies to both individual and collective intelligence systems (including human-AI teams), and becomes increasingly actionable as we learn more about the value of intuition and reflection.

Ulrike Hahn boosted

Milano University Press

@milanoup@mastodon.uno · last month

#newissue
Now online #TheReasoner's special issue about the inaugural meeting of the “Milan Logic and Philosophy of Science Network", held at Politecnico di #Milano in March 2025.

Part 1: #AI, #Robotics, and #Cognition
Part 2: #Epistemology, #Reasoning, and #Logic
Part 3: #Philosophy of the #PhysicalSciences, #Biology, and #Health.
Part 4: Science, Values, and Uncertainty.
Discussions: #gender, #mentalhealth and #publicengagement in academia

⬇️ https://riviste.unimi.it/index.php/thereasoner/issue/view/2713?mtm_campaign=mastodon

@cultura @scienze

Copertina the reasoner vol. 19, issue 4, decembre 2025

Nick Byrd, Ph.D.

@ByrdNick@nerdculture.de · last month

#AI may exacerbate bad academic habits:

Science summaries from large language models ( #LLMs) were nearly five times more likely than human-authored summaries to contain broad generalizations (95% CI [3.06, 7.70], p < 0.001).

And newer language models over-generalized more; not less!

https://doi.org/10.1098/rsos.241776

#sciComm #philSci #metascience #logic #edu #academia

Generalization bias in large language model summarization of scientific research

Konrad Hinsen boosted

Jencel Panic

@abuseofnotation@mathstodon.xyz · last month

In my articles about logic and thinking many people respond that "this is bullshit, logic is just a formal science, it isn't related at all e.g. with ideology" which is hard to disprove, but very wrong: In thinking, things are related in ways that are very subtle and hard to explain. And absolutely everything is related to ideology! e.g. art is related to ideology, religion is related to ideology (it can prompt you both to your stuff out and to being a nazi (two very different kind of beliefs)).

This is very important to understand, because many people already know it.

Why was the nazi propaganda so successful? Because they understood that just straight up telling people what to think, is not the only way to influence their opinion. So they didn't only create nazi politics, they created nazi art, nazi architecture, nazi science, nazi philosophy etc.

https://abuseofnotation.github.io/boolean-thinking/

#logic #politics #ideology

Abuse of Notation - writings on math, logic, philosophy and art - The case against boolean logic

⁂

More from

Jencel Panic

Milano University Press

@milanoup@mastodon.uno · last month

#newissue
Now online #TheReasoner's special issue about the inaugural meeting of the “Milan Logic and Philosophy of Science Network", held at Politecnico di #Milano in March 2025.

⬇️ https://riviste.unimi.it/index.php/thereasoner/issue/view/2713?mtm_campaign=mastodon

@cultura @scienze

Open Science Community Twente boosted

Gabriel Heinrichs

@gabrielheinrichs@akademienl.social · 2 months ago

Yay for the editors of Journal of Philosophical Logic for collectively quitting Springer & starting something better!

And yay for @openlibhums for funding this new-but-venerable diamond #openaccess journal.

"All of the editors-in-chief and associate editors of Springer Nature’s Journal of Philosophical Logic have announced their immediate or pending resignation and the launch, as of today, of a new “diamond” open access journal, Philosophical Logic."

https://dailynous.com/2025/12/10/editors-at-springers-journal-of-philosophical-logic-resign-launch-new-open-access-journal/
#philosophy #logic

Djoerd Hiemstra 🍉 and 1 other boosted

Tom de Jong

@de_Jong_Tom@mathstodon.xyz · 2 months ago

I'm happy that my proposal for an introductory course on Homotopy Type Theory / Univalent Foundations at the European Summer School in Logic, Language and Information 2026 in Prague was accepted!

A great opportunity to make use of @egbertrijke's recently published book, @MartinEscardo's lecture notes (in Agda!) and @danielgratzer and @carloangiuli's book draft!

Links for the curious:
- Egbert's book: https://doi.org/10.1017/9781108933568 & https://arxiv.org/abs/2212.11082
- Martín's notes: https://cs.bham.ac.uk/~mhe/HoTT-UF-in-Agda-Lecture-Notes/index.html
- Daniel and Carlo's book draft: https://www.danielgratzer.com/papers/type-theory-book.pdf

#logic #typetheory

Gabriel Heinrichs

@gabrielheinrichs@akademienl.social · 2 months ago

Yay for the editors of Journal of Philosophical Logic for collectively quitting Springer & starting something better!

And yay for @openlibhums for funding this new-but-venerable diamond #openaccess journal.

https://dailynous.com/2025/12/10/editors-at-springers-journal-of-philosophical-logic-resign-launch-new-open-access-journal/
#philosophy #logic

Tom de Jong

@de_Jong_Tom@mathstodon.xyz · 2 months ago

I'm happy that my proposal for an introductory course on Homotopy Type Theory / Univalent Foundations at the European Summer School in Logic, Language and Information 2026 in Prague was accepted!

A great opportunity to make use of @egbertrijke's recently published book, @MartinEscardo's lecture notes (in Agda!) and @danielgratzer and @carloangiuli's book draft!

#logic #typetheory

Nick Byrd, Ph.D.

@ByrdNick@nerdculture.de · 2 months ago

...we know that giving people various cues can make up for their poor understanding of the #logic of conditionals.

For instance, adding a scenario to a familiar conditional makes people far more likely to realize how others would test it...if it's a rule: https://dx.doi.org/10.3791/67794