Twitxy Saved

Anders Giovanni Møller

@AndersGiovanni

PhD student in @nerdsitu at @ITUkbh. Mostly doing CSS, NLP and other random AI things 🤖🙌🏼🔮
Copenhagen andersgiovanni.com Joined July 2013
610 Following 287 Followers 169 Posts

Latest posts 5

Anders Giovanni Møller @AndersGiovanni
Tomorrow (Oct. 22nd) starts the Danish Digitalization, Data Science and AI 2.0 (D3A) Conference in Nyborg 🇩🇰 The @nerdsitu group has organized a nice little workshop - “Networks, data, society, and AI”.

🤷🏼‍♂️ Why go there?

We have a great lineup of speakers talking about intersting things. Simple.

1️⃣ Incredible data and AI journalist at @version2dk, Therese Moreau (@ThereseMoreau) - if you follow AI in Denmark, you’ve probably read something from her or maybe seen her on DR Deadline.

2️⃣ Associate professor Pantelis Pipergias Analytis from @SyddanskUni, who does research within mathematical and computational models to study individual and aggregate human behavior.

3️⃣ CEO and founder of @ElectricityMaps, Olivier Corradi (@corradio). Electricity Maps provides real-time data and forecasting of electricity signals to enable data-driven solutions to decarbonize electricity consumption.

After individual presentations, we will have a panel discussion moderated by Vedran Sekara (@vedransekara).

🙋🏼‍♀️ Who is this for?

This workshop is for practitioners, policymakers, researchers, and everyone who loves to see data science applied to real-world problems, has an interest in computational social science, networks, AI, and wants to understand social and human behavior.

⏰ When and where?

Tuesday, October 22, 13:30 - Sal D.

Come check it out and say hi!🤙🏼
Anders Giovanni Møller @AndersGiovanni
👩🏼‍💻 Real or Robotic? 🤖 Can LLMs accurately simulate qualities of human responses in dialogue?

Human conversations with LLMs are great for assessing the capabilities of LLMs. But having lots of folks chat with LLMs is challenging (💰⏳🕵️). Could we have another LLM *simulate* being a human talking to an LLM as a substitute?

In our new preprint, we test whether models can roleplay as the human in human-LLM conversations. Using the WildChat dataset and 100K+ simulations we test how well these LLM responses actually mimic with human ones. Our study spans 🇬🇧 English, 🇨🇳 Chinese, and 🇷🇺 Russian, using 21 linguistic metrics like lexical, semantic, syntactic, and stylistic features.
3 replies 11 reposts 34 likes 7.1K views View on X