Eli Fennell, Ph.D.

Delving into LLM-assisted writing in biomedical publications through excess vocabulary Excess words track LLM usage in biomedical publications.

An analysis of abstracts of biomedical research published before and after the release of ChatGPT reveals that AI is having an unprecedented impact on scientific writing, with at least 13.5% of abstracts processed with LLMs, with some subcorpora reaching 40%.

#AIinScience #LargeLanguageModels

Problematic porn use remains stable over time and is strongly linked to mental distress, study finds A yearlong study of more than 4,000 U.S. adults found that problematic pornography use tends to persist over time and is strongly associated with higher levels of anxiety and depression, suggesting a…

New research published in Addictive Behaviors reveals that problematic porn use is correlated with greater psychological distress and tends to remain stable over time. Further research is still needed to clarify the causal relationships between variables.

#ClinicalPsychology #Addiction

Partisan identity drives social polarization more than race or religion, study finds A new study reveals that political party affiliation is the most powerful driver of social polarization in the United States—surpassing race, religion, income, and education

"[T]he data revealed that out-group hostility was more intense than in-group loyalty. Americans appear more motivated by whom they dislike than whom they support. Democrats displayed particularly strong negative feelings toward Republicans compared to the reverse."

#PoliticalPsychology

Trait Mindfulness and Prosocial Behavior: The Moderating Role of Self-Construals and Individualism - Mindfulness Objectives Trait mindfulness is associated with many measures of individual well-being, but its relationship to prosocial behavior is less clear. Prior research found that a brief intervention…

Newly published research has found that interdependent self-construal and vertical and horizontal individualism moderate the association between trait mindfulness and prosocial helping, consistent with prior research on state mindfulness.

#Psychology #Mindfulness

I will never support any argument that scientists should not have open, honest, and uncensored discussions about the good, the bad, and the ugly when it comes to the scientific method. If we start doing that, the enemies of science have already won.

They would have found something to distort to suit their agenda no matter what. You can't stop people who are determined to misrepresent science by being really really careful with your words.

In other words, if you have a hypothesis, which by definition is a prediction, you should be able to specify that before the research, but if you don't, you can just pre-register it as exploratory. There's no rule that says you have to have specific hypotheses before conducting research.

Pre-registration doesn't negatively impact discovery. You're still allowed to do exploratory research. All pre-registration does is limit the ability to fit a post hoc "hypothesis" to pre-existing data.

I never said that, so now you're putting words into my mouth.

I was literally responding to the exact points you brought up, which is the very definition of 'on track'. Sounds like you just want to make assertions and not have them challenged, even in a nuanced fashion which is not a very scientific attitude. But thank you for clarifying that.

To your point about the initial hyperbole, sure, I think some hyperbolic ideas were floated, but also that's a healthy part of having open discourse about how to solve problems. Sometimes an idea has to be discussed openly before the flaws become obvious. Condemning that is counterproductive.

All excellent points. I think the latter is in many ways the strongest evidence of the flaws in this argument, even if the most correlational in some ways. This comes off as the mentality of someone trying to figure out what they can do different not to "provoke" an abuser.

4) Replication at different labs before first publication. - Again, this is pretty extreme, though there are many resources now to make it easier to collaborate on large multi-site studies. Still, I think it would be sufficient to strongly note the lack of findings from other labs as a limitation.

3) 250 Person Samples - This is just bad statistics. The size you need for a sample depends on factors like the alpha and beta levels, the minimum effect size you want to detect, and the populations you want to infer to from your results (e.g. all humans = larger sample than caucasian males).

2) Eliminate Journals - Why? That does sound like a very extreme suggestion. Peer review improves the quality of the work, or at least the writing, when done properly. However, some journals did bad things like not publishing null results or replication studies. Science reformers helped end that.

Registered clinical trials make positive findings vanish - Nature A study showing a fall in positive trial results after the roll-out of clinicaltrials.gov attracted much attention on social media.

Also, why would you not want to preregister? The time involved in doing so is fairly negligible, and the benefit is increased confidence that you didn't just gather some data, run infinite analyses until you found something statistically significant, and then formulated a post hoc hypothesis.

Registered clinical trials make positive findings vanish - Nature A study showing a fall in positive trial results after the roll-out of clinicaltrials.gov attracted much attention on social media.

I'll address these one at a time: 1) Preregistration - All Federal Clinical Trials have required preregistration since 2000 and it has been an absolute net positive, e.g., - www.nature.com/articles/524... (1)

I'd also like to know what specific examples of proposed solutions you think are not good. They seem pretty common sense to me. Pre-registration. Data transparency. Sample size, justification and the use of large samples. Considering effect sizes and not just statistical significance. etc.

If there are specific examples, then those examples should be given. Simply saying that somehow the rhetoric was not good and has to change is useless. That's not science. Science is providing specific evidence and linking any proposed solutions to that evidence.

Heck, even as far as correlational evidence goes, this is weak. This isn't even time series data showing that the adoption of rhetorical reforms aligned with some significant increase in the acceptance of climate science.

Yes but you've made a specific claim that this has something to do with climate scientists "changing their rhetoric". That's a causal claim. All this shows is that climate science has been overall effective. Correlation doesn't equal causation.

No one has suggested that science reform is going to be 100% successful. You're arguing against a Straw Man.

And did that work? Where are the results showing that such efforts actually paid off?

There will be many victims of the regime undermining confidence in science. Scientists themselves. People (especially children) who will die of vaccine-preventable illnesses. The entire planet is suffering the effects of preventable/reversible climate change. etc. These are all linked together.

I agree. I would like to see specific examples where that occurred. Because this article cites no such examples, it just fixates on the use of the term "crisis".

Sure, but the article presents zero evidence that the science reformers didn't communicate in a careful and nuanced way. It seems narrowly fixated on the use of the term "crisis". But there was, ironically, no nuanced weighing of the case for and against using that term.

Basically, this meme is 100% accurate. There is no amount of careful you can be with science communication that can't be misrepresented by someone determined enough to do so. Using a regime with an obvious agenda to undermine science as an exemplar is frankly a bad faith argument.

Also, re:geneticists, you can prove in 10-minutes by searching YouTube that their efforts were in vain. Racists just ignore the things that don't suit their predetermined viewpoint, no matter where you place them.

Yes, I read the article, and it's nonsense. Regimes that want to undermine confidence in science have never had any difficulty doing so. If anything, what makes it easier is when sloppy research practices (e.g., p-hacking, inferring from underpowered samples, etc.) undermine the quality of findings.

In the context of what the article is referring to, yes, they would be victims of an Administration that is assaulting science.

Science needs to be brutally honest internally when it needs major reform, as it did. Opportunistic politicians will always find a way to attack us anyways.