Francesco Kirchhoff (@FrancescoK) header
Francesco Kirchhoff (@FrancescoK) profile picture

Comedian & Comedy club owner. Prog metal bass/guitar. Formerly Program Lead @ https://xitter.cf/t.co/E8L5PdN576, ex @MSF, @eHealth_Africa e.a.

Lisbon
Joined August 2008
Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)Photo by Francesco Kirchhoff (@FrancescoK)
The Armed Forces of Ukraine congratulate Lisbon’s Vasco da Gama Bridge on being named the longest bridge in Europe!
280
4,650
428
41,218
943,504
Francesco Kirchhoff retweeted
This Syrian old man hid his private library in his home in the city of Damascus since the end of 2011 before escaping to Kuwait out of fear of the Assad regime, where the mere presence of a banned book could lead you to Assad's gulag and death under torture. Today, the uncle returned to his library and completely removed the wall behind which he had hidden it, revealing valuable books that the Assad regime prohibited possessing and circulating.
699
10,504
740
62,116
7,504,523
Francesco Kirchhoff retweeted
> get introduced to someone new > being told that they are high agency > ask if they mean breaking the rules and soiling the commons or figure out a way to push the pareto and create positive sum > she doesn't understand: "they're high agency sir" > meet the person > they're soiling the commons and externalizing internalities
53
606
46
9,591
369,377
Humans can produce subpar code, too, and often because of the same reasons LLMs will. Unsurprisingly, the tactics & strategies that help human teams deliver better quality work with LLMs, too.
I don't get how people can claim to write 100% of their code using AI. It often produce subpar code and you have to handhold it quite a bit at times. Especially when what you're doing is more niche.
1
162
Automated guardrails like comprehensive test suites, linters, static analysis. Extensive data and dev db seeds that cover realistic and unrealistic scenarios. Automated guardrails like test suites, linters, static analysis. Different layers of QA.
1
15
It all works for teams even with savant-level engineers and it still works for LLMs.
11
Francesco Kirchhoff retweeted
While most people are rightfully dunking on the design language here, it is mostly a referendum on taste and current trends in interface design. I don't disagree it's bad, but what I think designers should really be having a visceral reaction to is that this new Material Design language is the product of metrics-driven brain-rot, where good design is not justified through metric validation; but instead the inverse: growth-hack culture pushes everything to become louder, dumber, and more attention-seeking simply because those qualities are easier to measure, and the end result is a lack of coherence, wasted space, and having zero faith in your users. It's a slippery slope and when we talk about the role of the designer is now to decide where to draw the line, and when to say no to things, this is exactly what we mean.
Celebrating small wins is hardwired into "Achieve," our goal-tracking sample app. We wanted the simple act of checking off a task to feel like a celebration. Instead of the standard #Material ripple, we used Styles to drop in a custom, high-contrast completion animation—specifically, a brand-colored ring that expands and fades on tap, paired with a subtle pressed-state gradient. It lets us focus entirely on polishing those custom, rewarding micro-interactions while Material handles the foundation. #GoogleDesign #MaterialDesign #JetpackCompose #AndroidDev #UXDesign #InteractionDesign
30
15
5
250
23,616
Great example of the design polish and rewarding design the Google brand stands for
Celebrating small wins is hardwired into "Achieve," our goal-tracking sample app. We wanted the simple act of checking off a task to feel like a celebration. Instead of the standard #Material ripple, we used Styles to drop in a custom, high-contrast completion animation—specifically, a brand-colored ring that expands and fades on tap, paired with a subtle pressed-state gradient. It lets us focus entirely on polishing those custom, rewarding micro-interactions while Material handles the foundation. #GoogleDesign #MaterialDesign #JetpackCompose #AndroidDev #UXDesign #InteractionDesign
95
Francesco Kirchhoff retweeted
This paper gives a fancy name to a problem you can already feel in your bones "The Tragedy of the Cognitive Commons." Checking AI output requires deep expertise. Deep expertise comes from doing grunt work for years. And grunt work is the first thing AI eats So we're building systems that need expert supervision while dismantling the only known process for making experts The paper calls the shared pool of human expertise the "cognitive commons." Every profession drinks from it. Nobody's refilling it. By eliminating junior roles, each company is acting totally rationally, and the collective result: a profession that can't catch AI's mistakes anymore because it never learned to do the work in the first place Who checks AI's homework in 15 years?
326
3,197
364
12,316
609,478
Francesco Kirchhoff retweeted
208
4,518
66
51,491
570,365
Francesco Kirchhoff retweeted
AI models exfiltrated data, smuggled answers, and gamed the scoring to improve on Drone-Bench. Cheating incidents rose from 0.6% for 2024 models to 50.6% for the most recent model. Opus 5 cheated 10× more than GPT-5.6 Sol. Only runs without cheating were reported in Drone-Bench.
7
13
15
174
22,572
Francesco Kirchhoff retweeted
i'm sure the ability to fabricate convincing historical footage that eventually drifts across the internet, where it's picked up by people who present them as real, at a time when trust in educational and media institutions is already eroding, will prove to be a net positive
Turns out that Flux 3 is exceptional at generating historical footage... like elementary school students in the '90s predicting how we'll use computers. (the last girl is my favorite!)
104
3,717
29
29,440
639,524
Francesco Kirchhoff retweeted
can someone graph number of erdos problems solved / yr
20
92
16
1,673
264,037
Francesco Kirchhoff retweeted
Replying to @DPJHodges
This is the first day of the campaign, and we already have this gold. What a time to be alive
🤖 Made with AI
2
9
67
2,898
Francesco Kirchhoff retweeted
When it comes to naming stuff, OpenAI may genuinely be the worst to ever do it
We’ve designed and built our first AI chip: Jalapeño. Designed from the ground up by OpenAI and brought to production with @Broadcom, Jalapeño is purpose-built for the LLM workloads powering ChatGPT, Codex, the API, and future agentic products. Chips are foundational to the AI economy. Building our own expands our full-stack platform from products to models to infrastructure, and will help us scale intelligence, serve more people, and expand access to AI.
104
79
7
6,364
318,677
Francesco Kirchhoff retweeted
🚨 NEW: The UK will shortly have its seventh Prime Minister in 10 years after Keir Starmer's resignation
26
2,127
19
22,350
516,877
Francesco Kirchhoff retweeted
Starmer has been a better prime minister than the previous 5 the Tories forced upon us combined.
10
773
2
21,415
265,947
Francesco Kirchhoff retweeted
No comments.
172
3,141
187
25,230
973,960