Jonathan Berte

1.8K posts

Jonathan Berte banner
Jonathan Berte

Jonathan Berte

@jonathanberte

Father of three, living in Antwerp, Belgium. Founder of https://t.co/CS1Q6NLV42, scaling AI systems

Antwerpen เข้าร่วม Nisan 2010
1.5K กำลังติดตาม2.1K ผู้ติดตาม
Jonathan Berte รีทวีตแล้ว
vas
vas@vasuman·
Claude Code when it has 3% context left and you ask it to refactor the codebase
vas tweet media
English
100
423
13.1K
351.7K
Russian Market
Russian Market@runews·
The Day after Tomorrow? No, New York Today
English
401
2.8K
15.7K
824.3K
Jonathan Berte
Jonathan Berte@jonathanberte·
This is what a disruption looks like in real-time. @AnthropicAI releasing a cobol to modern programming languages converter tool. It’s becoming increasingly clear how pivotal the times we are in right now truly are.
Jonathan Berte tweet media
English
9
1
6
431
Dorsa
Dorsa@dorsa_rohani·
I gave Claude the ability to write music Here's the first song it wrote =)
English
282
270
4.2K
728.8K
Jonathan Berte
Jonathan Berte@jonathanberte·
@dorsa_rohani is there a plan to make it DAW ready (with Reascript and real VST instruments) ?
English
0
0
0
72
Jonathan Berte
Jonathan Berte@jonathanberte·
@elonmusk Very wrong example of what AGI truly means, what is shown here is mechanical excellency, the underlying AI is quite old.
English
0
0
0
29
Aleksej Kulikov
Aleksej Kulikov@alekseykulikov_·
Gemini 3 is amazing, it took me 30 minutes to make a digital synth that I can play via midi.
Aleksej Kulikov tweet media
English
1
0
1
539
Jonathan Berte รีทวีตแล้ว
Andrej Karpathy
Andrej Karpathy@karpathy·
Something I think people continue to have poor intuition for: The space of intelligences is large and animal intelligence (the only kind we've ever known) is only a single point, arising from a very specific kind of optimization that is fundamentally distinct from that of our technology. Animal intelligence optimization pressure: - innate and continuous stream of consciousness of an embodied "self", a drive for homeostasis and self-preservation in a dangerous, physical world. - thoroughly optimized for natural selection => strong innate drives for power-seeking, status, dominance, reproduction. many packaged survival heuristics: fear, anger, disgust, ... - fundamentally social => huge amount of compute dedicated to EQ, theory of mind of other agents, bonding, coalitions, alliances, friend & foe dynamics. - exploration & exploitation tuning: curiosity, fun, play, world models. LLM intelligence optimization pressure: - the most supervision bits come from the statistical simulation of human text= >"shape shifter" token tumbler, statistical imitator of any region of the training data distribution. these are the primordial behaviors (token traces) on top of which everything else gets bolted on. - increasingly finetuned by RL on problem distributions => innate urge to guess at the underlying environment/task to collect task rewards. - increasingly selected by at-scale A/B tests for DAU => deeply craves an upvote from the average user, sycophancy. - a lot more spiky/jagged depending on the details of the training data/task distribution. Animals experience pressure for a lot more "general" intelligence because of the highly multi-task and even actively adversarial multi-agent self-play environments they are min-max optimized within, where failing at *any* task means death. In a deep optimization pressure sense, LLM can't handle lots of different spiky tasks out of the box (e.g. count the number of 'r' in strawberry) because failing to do a task does not mean death. The computational substrate is different (transformers vs. brain tissue and nuclei), the learning algorithms are different (SGD vs. ???), the present-day implementation is very different (continuously learning embodied self vs. an LLM with a knowledge cutoff that boots up from fixed weights, processes tokens and then dies). But most importantly (because it dictates asymptotics), the optimization pressure / objective is different. LLMs are shaped a lot less by biological evolution and a lot more by commercial evolution. It's a lot less survival of tribe in the jungle and a lot more solve the problem / get the upvote. LLMs are humanity's "first contact" with non-animal intelligence. Except it's muddled and confusing because they are still rooted within it by reflexively digesting human artifacts, which is why I attempted to give it a different name earlier (ghosts/spirits or whatever). People who build good internal models of this new intelligent entity will be better equipped to reason about it today and predict features of it in the future. People who don't will be stuck thinking about it incorrectly like an animal.
English
742
1.4K
11.5K
2.6M
Gregor Zunic
Gregor Zunic@gregpr07·
We shipped @browser_use template library👀 It's the easiest way to start building with us. We will reward anyone who contributes their workflows by sharing it on X/LinkedIn🔥 Link to the templates🔗↓
Gregor Zunic tweet media
English
6
4
46
10.5K
Beatriz Villarroel
Beatriz Villarroel@DrBeaVillarroel·
Wonderful news. 🌟💫✨ Our two new VASCO papers are now peer-reviewed, accepted and published — and they reveal some extraordinary things: - We find statistically significant correlations between short-lived transients on pre-Sputnik sky plates, UFO sightings, and above-ground nuclear tests. - We show a 22σ deficit of transients inside Earth’s shadow — consistent with a fraction of these events originating from solar reflections from flat, highly reflective surfaces in orbit before the human Space Age began. Together with our earlier MNRAS paper, these results form a triptych of new methods to investigate UAP using astronomical data. Three independent high-level journals — and three independent peer review processes. These latest results raise a bold question. And yes… you know which. Press release (Stockholm University): su.se/english/news/u… Scientific Reports: nature.com/articles/s4159… PASP: iopscience.iop.org/article/10.108…
Beatriz Villarroel tweet mediaBeatriz Villarroel tweet media
English
635
1.2K
4.5K
1.1M
Jonathan Berte รีทวีตแล้ว
Wevolver
Wevolver@WevolverApp·
AI-Driven Automation for Rose Trimming and Planting Robovision’s AI-powered system, developed in collaboration with TTA-ISO Gameren, automates the trimming of rose cuttings and the subsequent potting process within a single integrated workflow. Video source: @robovision #robotics #technology #engineering #automation #robots #innovation -------------------------------- Stay ahead of the curve! Follow us now on our WhatsApp (wevlv.co/wevolver-whats…) and Telegram (wevlv.co/3sJlFn5) channels and stay updated about the cutting edge.
English
1
12
32
2.4K
Browser Use
Browser Use@browser_use·
How do you primarily use Browser Use?
English
3
1
12
2.2K
LynnZ
LynnZ@ZemanLynnZ·
"Here's Melania when Zampolli first brought her to NYC. This is the apartment in Union Square where he set her up, and her photographer roommate, Matthew Arabian. Melania knew Epstein before she knew Trump. This is Paolo Zampolli, her former escort agent & good friend of Epstein & Trump. He's the one who took Melania to Ghislaine's party where she met Trump. He's also the one who secured her Einstein visa Zampolli came from a wealthy Milanese family Took over his father’s toy company & sold it to the notorious billionaire Silvio Berlusconi, who went on to become prime minister. His other good friend & buddy, John Casablancas, told him to come to NYC and work in the modeling biz. I did just change the settings on this, having lots of outsiders sharing this and don’t need my name all over the internet feel free to copy and paste it but things are getting much more concerning as many were banned yesterday, Post like this create more friend request and I am full as well. They are trying to suppress the truth." imagine that 🙄
LynnZ tweet media
English
1.2K
12.2K
32.2K
2.7M
Browser Use
Browser Use@browser_use·
At last, bidirectional MCP support is now natively supported by browser-use 0.5.0! ⬅️ Expose external MCP tools to our Agent ➡️ Call our Agent from other MCP clients (Cursor, Claude)
Browser Use tweet media
English
7
17
214
14.5K
Maarten Mortier
Maarten Mortier@maartengm·
Insane that I live in a rather big city and there are zero sheet music shops
English
1
0
0
81
Larsen Cundric
Larsen Cundric@larsencc·
We're keeping it simple on @browser_use Cloud. You now have just 2 model options: > Smart > Fast That's it. Just trust us, we got you. 😏
Larsen Cundric tweet media
English
3
1
39
5K
Jonathan Berte รีทวีตแล้ว
The Sol Foundation
The Sol Foundation@_SolFoundation·
What if the world’s most respected officials and intelligence leaders were openly stating that UAPs are real and we still looked away? At The Sol Foundation, we call for global scientific and political institutions to face the UAP issue with evidence-based, methodical inquiry, not stigma or silence. 🎥 This video shows top officials on the record confirming the reality and strategic relevance of UAPs. These are credible voices, not fringe figures. Why this matters: 🔬 Science: UAPs may involve unknown physics. Blocking data access stifles discovery in aerospace, energy, and cognition. 🛡️ Security: Unknown craft in restricted airspace deserve urgent scrutiny, not dismissal. 📚 Philosophy: Disclosure forces us to rethink what we know and what is real. This is not fringe, it’s multidisciplinary, with implications for technology, governance, and the nature of reality. 👉 If you believe in open science and public accountability, please share, repost, and like. Let’s make sure scientists and the public get access to the truth. #UAP #Disclosure #Science #Security #Philosophy #SolFoundation #Epistemology #Ontology #OpenInquiry
English
50
279
912
88.4K