These 10 years on Twitter have been an amazing adventure.
Thanks for making it so special and so intriguing!
I am taking a break, the recommender system here has started to profile and pollute lines in ways that I do not like.
Reach me here: bsky.app/profile/massimost.b…
Happy to announce the CogNosco & Ascolto podcast, featuring our research papers in #cognitive network science and #AI. The first 4 episodes are out on YouTube and discuss cognitive networks in relationship with STEM Education, social media and creativity:
youtube.com/watch?v=vbtIrrvf…
Are you interested in AI safety… but you have a different background and are not sure of the best way to get into it?
Then this fellowship is precisely for you!
pibbss.ai/fellowship/
Happy 2025! 🎉
This will be the year my PENSO project starts, with a funding of 1.3 #million EUR from Fondo Italiano per la Scienza (FIS2).
As the only PI, my team will investigate the psychological implications of how #LLMs can convince, support and teach to humans. So exciting!
Happy 2025! 🎉
This will be the year my PENSO project starts, with a funding of 1.3 #million EUR from Fondo Italiano per la Scienza (FIS2).
As the only PI, my team will investigate the psychological implications of how #LLMs can convince, support and teach to humans. So exciting!
The project starts in March 2025 and I will have to hire several postdocs, PhD students and a Lab manager.
Follow updates and job offers through our CogNosco Lab website:
cognosco.dipsco.unitn.it/
By mapping the meanings of the words used to communicate emotions across more than one-third of the planet’s spoken languages, a study in Science found that there is significant variation in how emotions are expressed across cultures. #ScienceMagArchivesscim.ag/4gGECui
How creative are large language models compared to humans? This paper tested 13 standardized creativity tasks, finding LLMs rank at the 52nd percentile overall - essentially average at ideation tasks but notably weaker at creative writing. arxiv.org/abs/2412.03151
Meta-analysis of PISA creative thinking data by @BenjaminGoecke across 62 countries (n=493,660) finds SES is a strong predictor of creativity, along with gender: girls scored higher than boys—a bigger gender gap than boys’ slight advantage in math. osf.io/preprints/psyarxiv/zy…
Excited to share @CognoscoL´s new preprint on creativity! 😍Using textual forma mentis networks and explainable AI to analyse ~1000 human and GPT-generated stories to predict creativity ratings. Thanks to my co-authors Natalie Fischer @dsalvaz@GiulioRossetti@MassimoSt!
ALT Feature importance across the three models: humans-rating-humans, GPT-rating-humans, GPT-rating-GPT.
Class 0 = low creativity stories
Class 1 = mid creativity stories
Class 2 = high creativity stories
😍Announcing LWOW - the LLM World of Words data repository for free associations, by @katie_abramski (lead author), Riccardo Improta, @GiulioRossetti and myself (co-last w. Rossetti).
github.com/LLMWorldOfWords/L…
With 3 mil. responses for free associations from Mistral, Llama3 ...
...and Haiku, #LWOW mirrors the SWOW data repository, coordinated by S. De Deyne @5i111011 and other international researchers.
👇
In the presenting white paper here on ArXiv - arxiv.org/abs/2412.01330 - we show how free associations can be used to highlight #gender#bias.
The platfom will keep growing in the future as we are working on massive simulations of larger #LLMs. Send us an email if you want to get in touch and/or provide feedback on the preprint!
@UniTrentoDIPSCO @kdd_lab@IstiCnr_It@Unipisa