Julian Ustiyanovych

Passioned about AI, Behavioural (Data) Science, and LLMs Psychology.

[Kai-Fu Lee @kaifulee]

The biggest revelation from Deepseek is that Open Source has won. For a 1% difference in performance, it will be difficult for OpenAI to justify its price when the competition is free and formidable. -from my interview with Bloomberg

[Julian Ustiyanovych @j16h0]

@elonmusk Omg, you are such a stupid dick head, Elon!

[Julian Ustiyanovych @j16h0]

@elonmusk Consumer sentiment plays a significant role in market dynamics, and history has shown that public perception can make or break even the most successful brands.

[Julian Ustiyanovych @j16h0]

@elonmusk it's important to recognize that branding plays a crucial role in public perception. Musk has made himself the face of Tesla, meaning his actions and statements directly influence how people feel about the company.

[Julian Ustiyanovych @j16h0]

@elonmusk Brand reputation is fragile. If Musk continues to act in ways that alienate customers or investors, it’s not surprising that Tesla, Starlink, Twitter, and his other ventures could face backlash.

[Julian Ustiyanovych @j16h0]

@elonmusk When a public figure engages in controversial or ethically questionable behaviour, such as allegedly leveraging Starlink access in Ukraine for geopolitical and economic gain it's understandable why some people would choose not to associate with his companies.

[Julian Ustiyanovych @j16h0]

@elonmusk, you spineless clown. Cutting off Starlink to Ukraine unless we hand over our resources? That’s not neutrality - it’s extortion. You’re just @DonaldJTrumpJr
lapdog now. Hope the paycheck’s worth the legacy you’re burning. Pathetic idiots.

https://t.co/nYVBoOjgG9

[Julian Ustiyanovych @j16h0]

#GPT4, with 600 human raters, assessed 226 public figures' personalities. Results? #GPT4 nailed it with correlations from r=.76 to .87, outperforming models built for this task 🤯 The kicker? It wasn’t even trained or given feedback! https://t.co/tvO5njkJ42

[Julian Ustiyanovych @j16h0]

8/8. To conclude, LLM-based simulations of experiments could offer significant value in areas such as (1) intervention design, (2) minimizing harm to human participants, (3) predicting subgroup effects, and (4) pre-testing product hypothesis, etc.

[Julian Ustiyanovych @j16h0]

#SyntheticUsers (or #HomoSilicus) - who are they?

#SyntheticUsers - are artificial #agents that can participate in survey, scenario, and simulation based behavioural research like humans.

[Julian Ustiyanovych @j16h0]

7/8. A recent study titled "Predicting Results of Social Science Experiments Using Large Language Models" found a strong alignment (r = .85) between simulated and observed effects across 70 pre-registered studies, 476 treatment effects, and over 100K participants.

[Julian Ustiyanovych @j16h0]

@leijun why it's not reflected on a stock market?

[Julian Ustiyanovych @j16h0]

@ProjectJupyter @deno_land @vega_vis @d3js_org @DataPolars Love it!

[Julian Ustiyanovych @j16h0]

@elonmusk What's the point to name it 𝕏, while you still have a button "Tweet"?