Ali O. Ilhan

The Last Human Job A timely and urgent argument for preserving the work that connects us in the age of automation"A compelling case for valuing care as a societal good and as skilled labor."—The Nation

Another good book.
press.princeton.edu/books/hardco...

Bacteria to AI A new theory of mind that includes nonhuman and artificial intelligences.   The much-lauded superiority of human intelligence has not prevented us from driving the planet into ecological disaster. For...

Although I have been training, fine tuning, using, playing etc. with ML models for research since roughly 2009, my understanding of AI in general has changed tremendously in last two months. This book is one of the reasons, highly recommended. press.uchicago.edu/ucp/books/bo...

These are taken from his tweet, not my words.

5. o3 proves that we needed a new architecture, not just scaling old models
6. o3 still fails on some very easy tasks
7. o3 should score in an upcoming benchmark under 30% (humans 95%)
8. We don't know if these capabilities will extend to other domains

1. o3 doesn't seem to be AGI, but it represents a significant leap forward
2. It's a genuine breakthrough in adaptability and generalization
3. o3 is capable of adapting to tasks it has never encountered before
4. This generality is too expensive, and it's not economically feasible today

One of the most level-headed analysis of o3, by François Chollet exlained by Santiago Valdarrama. Btw, Chollet is the creator of Keras, and one of the most knowledge people on the planet about deep-learning. x.com/svpino/statu...

homer simpson says okay im done now in front of a house ALT: homer simpson says okay im done now in front of a house

Done with grading!

Trying Llama and Mistral for sentiment analysis in a very weird child welfare dataset, the results are pretty promising (many other models were not as successful).

Crazy stuff

:)))

"Here are some ways of making your study replicable. (No, the first steps are not preregistration or increasing the sample size!)"

statmodeling.stat.columbia.edu/2023/06/22/h...

"I think there’s a problem with trying to fix the replication crisis using procedural reforms, by which I mean things like preregistration, p-value or Bayes-factor thresholds, and changes in the processes of scientific publication."

statmodeling.stat.columbia.edu/2024/09/26/w...

We are hiring two non tenure-track Assistant Professor-Educators at University of Cincinnati's Industrial Design Program.
jobs.uc.edu/job/Cincinna...

Ollama Get up and running with large language models.

I had no idea working with LLMs through ollama.com was that easy (I was doing it the hard way). +As you can imagine it has python integration. If you have a decent GPU you can play with the models locally or integrate them into your workflows.

A Review of The Quantified Scholar: How Research Evaluations Transformed the British Social Sciences | The Quantified Scholar: How Research Evaluations Transformed the British Social Sciences, Juan Pa...

I reviewed @pardoguerra.bsky.social 's excellent "The Quantified Scholar" for She Ji. It turned into a semi opinion piece about lack of evidence based discourse regarding research evaluation and its larger effects on design scholars and scholarship: www.sciencedirect.com/science/arti...

1st online training of 2024 done.

Latest example of mission creep: the Office of Research Integrity requires us to watch a video on the peer review process, w/ script written for - and by - someone who has never submitted, reviewed, or edited an article or book. Collectively, hours of wasted time.

I should also get my picture taken like that :)

Carl Bergstrom looking like some Ted-talk-giving-mf in Australia.

It gives me a measure of hope that a headset mic and black backdrop has so quickly become the universal signifier of a bullshit thought-leader with nothing of substance to say.

Free online workshop: Webscraping using R, Thursday, 30 November, 15:00–16:30 (Irish time)

Sarah King, PhD Researcher and member of text-and-policy.com, will be teaching a free online workshop on webscraping using R. The event is open to all scholars regardless of university affiliation.

Register here: forms.gle/Q3pixYrbrEY6...
More info: www.ucd.ie/connected_po...

What Happens to Illustrators When Robots Can Draw Robots? Illustrators have made a living creating book covers. Now, artificial intelligence is learning to make art.

Source: www.nytimes.com/2023/11/04/b...

“The funny thing is, in so much science fiction it was theorized that robots, A.I., all this stuff was going to take over the drudgery, the hard labor, and free up humans to do creative work,” Paolini said.“Instead the A.I. is taking over the creative work, and we’re all stuck doing the hard labor.”

Scientific reform, citation politics and the bureaucracy of oblivion Berna Devezer, Bart Penders Abstract Current reform movements in science seek to change how researchers do science, the tools and infrastructure they use to so, and how they assess each others’ work in terms of quality and value. Here, we argue that openness and replicability are quickly becoming key indicators for such quality assessments and they sometimes operate through citation strategies that actively pursue (some degree of) oblivion for non-reformed science. We do not oppose a genuine pursuit of transparency and methodological quality, but are concerned by how uncritical and oversimplified interpretations of both are skewing the collective memory of the scholarly community.

Happy to see @penders.bsky.social and I's short paper out at Quantitative Science Studies: "Scientific reform, citation politics and the bureaucracy of oblivion"

Highlighting two recent examples, we warn against distorting the scientific literature based on conflation of bureaucracy and quality.

Unpacking P-hacking and Publication Bias (November 2023) - We use unique data from journal submissions to identify and unpack publication bias and p-hacking. We find initial submissions display significant bunching, suggesting the distributi...

Very cool paper just out in AER. They actually find *negative* amounts of publication bias - review process selects *against* marginally sig. results, even though reviewers like significance. Observed bunching is driven by p-hacking before submission.
www.aeaweb.org/articles?id=...

Computational social scientists and other sociologists who play with or think about data:

UNC-CH Sociology is looking for an Assistant Professor of Sociology and Data Science.

Application review begins 11/3!

unc.peopleadmin.com/postings/268...

A screenshot of the title and abstract. Authors: Berna Devezer and Erkan O. Buzbas Urgent attention is needed to address generalizability problems in psychology. However, the current dominant paradigm centered on dichotomous results and rapid discoveries cannot provide the solution because of its theoretical inadequacies. We propose a paradigm shift toward a model-centric science, which provides the sophistication to understand the sources of generalizability and promote systematic exploration. In a model-centric paradigm, scientific activity involves iteratively building and refining theoretical, empirical, and statistical models that communicate with each other. This approach is transparent and efficient in addressing generalizability issues. We illustrate the nature of scientific activity in a model-centric system and its potential for advancing the field of psychology. Keywords: generalizability; modeling; model-centric; result-centric; exploration

Received a delightful email from a colleague this morning informing me that their phd student writing a #metasci dissertation was raving about one of our papers: "Rigorous exploration in a model-centric science via epistemic iteration"
I want to talk about it a bit.
psycnet.apa.org/record/2023-...

:=)

Sometimes it is terminal velocity, sometimes it is quantum superposition, sometimes it is Puss in Boots, sometimes it is Bastet incarnate but mostly it is people who still wanna play the engagement game after they ran out of steam with Covid BS.