Game director @ Wildlife Studios. Building games with AI, scaling studios, and writing Calibrating Posteriors on epistemology and the bugs in your reasoning.
Joined January 2021
- Tweets 3,979
- Following 754
- Followers 786
- Likes 21,745
311 Photos and videos
Pinned Tweet
13 Nov 2025
I'm a gamer who built mobile games that were downloaded 300M times, and now I'm helping other developers scale their games through Wildlife's publishing division.
I tweet about gaming, the mobile market, AI, economics, politics, technology, and memes
DMs open for partnerships!
7
4,340
Luiz Piccini retweeted
Jun 14
Rio may have found the most overqualified municipal employee in human history.
Jun 13
SITUATION DETECTED: The city of Rio de Janerio has post-trained a model.
Based on Qwen 7/2, Rio 3.5 Open 397B adds SwiReasoning on top of the base Qwen model — a framework that dynamically switches between standard chain-of-thought and latent-space reasoning, guided by entropy-based confidence signals, so the model only "thinks out loud" when it needs to and otherwise reasons silently in hidden space for better token efficiency.
22
198
5,466
338,747
0.3% of the water consumed by US golf courses last year
BREAKING: Amazon data centers used 2.5 billion gallons of water in 2025
Community note
Amazon's data centers withdrew 2.5 billion gallons of water in 2025, down 2% from 2024 despite expansion, or less than 0.1% of annual U.S. landscape irrigation. aboutamazon.com/news/sustainab…
433
2,669
42,876
5,987,978
Luiz Piccini retweeted
Jun 11
We’re teaming up @Palmeiras, the first football club to meaningfully build upon TacticAI: our AI system that can help simulate field scenarios and predict open play dynamics up to 8 seconds in advance. ⚽
115
374
3,293
993,971
Jun 10
I created a benchmark for comparing models capacity to make games! As expected, Fable 5 is the best in all dimensions, but surprisingly it was cheaper and faster than Opus 4.8! Despite costing more tokens, taking less retries to reach the solution paid off in the end.
1
1
5
126
Jun 3
"There's a reason no one who gets a license ever uses public transit or walks anywhere ever again if they dont have to"
What the actual fuck
Jun 1
Thats a horribly inefficient waste of time. You probably end up spending close to 5 hours just in grocery stores every week if you do this (nevermind the extra hours spent walking there and back as well), when you add up shopping waiting in line. I'd rather just do 2-3 weeks worth of shopping at once in <2hours, drive my groceries right to my front door, and then not have to go to a store again for a few weeks
There's a reason no one who gets a license ever uses public transit or walks anywhere ever again if they dont have to - its way worse in every way.
Society should be more accommodating to cars, not less. Bike lanes for example, should be illegal
64
May 28
No one knows how to make an app.
When we build with AI, we’re drawing on a civilization-sized stack of software, hardware, infrastructure, and accumulated knowledge that no single person understands end to end.
But that has been true of modern creation for a long time.
[1/2]
1
94
May 28
No one knows knew how to make apps in 2022 either. No one knows how to make a pencil alone, either.
We are always relying on a global network of people we don't know. AI only made this even easier.
I wrote a short essay about this:
open.substack.com/pub/calibr…
68
Luiz Piccini retweeted
May 24
Anthropic onboarding day: Michael Scott introducing Karpathy like he just signed Wemby in free agency.
396
1,486
17,621
2,351,096
Luiz Piccini retweeted
May 22
(1) We are likely on track to develop AI systems capable of causing human extinction/permanent disempowerment, quite possibly within the next few years
40
56
585
295,237
Luiz Piccini retweeted
May 22
One day AI will cure a disease. And many people in this platform will look you dead in the eye and say
“the cure was already in the corpus of human knowledge and it’s not truly intelligent”
184
93
1,601
33,063
May 19
POV: you have to look at one of my Merge Requests
1
59
May 19
O Brasil não é pobre à toa
May 17
Quanto menos Data Center no Brasil, melhor. Servem para drenar energia, água e gerar ainda mais dinheiro para bilionários. Apodrecem qualquer região onde são instalados
39
Luiz Piccini retweeted
May 18
"it's just a really cool search engine. Useful for some stuff."
Yeah, in the same way that fire is useful for some stuff.
May 17
Just a reminder: the AI industry is a bubble.
AI itself is not a bubble, it's just a really cool search engine. Useful for some stuff.
13
5
165
4,423
May 18
POV: You don't understand supply and demand
1
52
May 18
This explain why this user thinks AI is dumb
May 18
AI is just as dumb as its user.
4
64
May 17
Many such cases. In so many subjects.
This guy also often say "this should be taught in school!", even though he wouldn't have paid attention either
1
79
Luiz Piccini retweeted
May 12
Woah! This is blasphemy. Milton Friedman is correct about the following:
1. He proved that the Great Depression was due to Fed contracting money supply. Ben Bernanke conceded this on Friedman's 90th birthday!
2. He was the first (and correct) to critique the Phillips curve relationship between inflation and unemployment. Past governments had tried to exploit the relationship. But he demonstrated that doing so simply shifts expectations.
3. He was instrumental in arguing against the Bretton-Woods regime. He (correctly) predicted that the flexible exchange rate world (that we're living in now) is much more stable.
4. The EITC is simply an implementation of the negative income tax he proposed as a replacement to bureaucratic welfare.
5. He was also correct about corporate tax being a regressive tax. Decades later, virtually no economists disagree.
I could go on and on and on. Milton Friedman was one of the greatest geniuses. Saying he wasn't right says more about your understanding of economics than his.
Economics is not a natural science and Milton Friedman hasn’t been correct about anything meaningful to policy or the actual world
67
352
2,741
126,064
Luiz Piccini retweeted
Counterpoint: it rules that we created a godlike intelligence and it agrees with me on practically everything politically
it is actually worrying that the models seem to have converged on similar beliefs on all important questions. they’re are neobuddhist neolibs which talk about annata and housing policy, including grok and the Chinese models! boring
23
117
2,578
107,932
May 9
Pedi pro Claude fazer o mesmo o mesmo para o Brasil: achar 10 políticas simples que melhorariam significativamente o país
Asked Claude:
'There's a meme called the "fix everything easily switch". What policies do you think are the best candidates for being a real fix everything switch in the US? Give me your top ten, your confidence, your reasoning, and why a given policy has not been implemented.'
2
4
249
May 9
9. Lista aberta é o pior sistema eleitoral do mundo democrático. Cria 33 partidos, deputado paraquedas, caixa 1 estrutural. Distrital misto alemão, cláusula de barreira de 5%, fim do fundão. Sem reformar como elege, nenhuma outra reforma passa. É a meta-alavanca.
1
2
73
May 9
10. O Brasil tem 140 estatais federais. Correios perde dinheiro num mercado onde Mercado Livre lucra. BB compete com Itaú sem mandato público claro. Saneamento federal atende metade do país e mata gente de diarreia. Privatizar não é ideologia — é parar de queimar capital.
2
75