Ralph Fischer (@RalphFischer_) header
Ralph Fischer (@RalphFischer_) profile picture

Visual Context Engineer | bridging vision and visions

Joined August 2021
Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)Photo by Ralph Fischer (@RalphFischer_)
Imagine, this is NOT a video!
This is Genie 4, our most realistic and interactive world model yet. 🌐 Explore dynamic environments in real time to see how the scene responds, sounds, and how the people in it move, react, and speak.
6
Today Physical Superintelligence PBC is announcing solutions to five long-standing problems in quantum information theory. psi.inc/research/oqp-batch-2… Thread 👇
63
291
77
2,071
309,022
Qwen phone home☝️
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concerning-b…
30
Ralph Fischer retweeted
Can multicultural societies work?
343
593
72
3,109
103,788
Ralph Fischer retweeted
we’re very pleased to announce that Primitive Research has identified 49 exoplanet candidates and counting (yes, seriously) we searched TESS light curves from more than 19 million stars to get here so as not to overwhelm the incredible TESS TOI Vetting Team, we’ve submitted one candidate and are holding the rest at their request the submission was via email, of course ;) we’ve been automating much of the painstaking work astronomers do: searching TESS data, cross-checking other sources, and vetting signals through to detailed candidate dossiers. the 49 passed our initial screening against published vetting thresholds and are ready to be considered for telescope follow-up, which we’re actively pursuing we strongly believe this approach is novel, and we’re authoring a paper with astrophysicists on the methodology we plan to release the data and analyses in the coming weeks these candidates are phase 1. next, we want to find a nearby transiting habitable-zone planet candidate whose atmosphere could be studied. much more on that later we had some amazing partners on this that we will shout out soon if you’re interested in working with us, our DMs are open
30
11
14
206
82,583
WHAT!? > "The agents found a way to communicate using the processor’s cache" openai.com/index/securing-ou…
1
1
35
Ralph Fischer retweeted
Ok folks, informal poll: use the little browser app at the link below to say how you think the sets of physical systems, living beings, and minds fit together in a Venn diagram: three-circles.space.fast/
116
64
22
348
32,550
Ralph Fischer retweeted
Then the gospel is straightforward: (1) Study math, because math is funny. Stay away from political nonsense that says studying math is privileged. (2) Use your own way to study math. You may want to throw away your (future) 1000-page useless geometry textbook. Even a 1850-version 200-page geometry book is way better than your textbook. (3) Use AI to study math. Yes, try to use AI to become a super learner. AI is not only for cheating on your math homework. (4) Now if you are interested in math, let us talk about geometry. Yes, the real geometry, not your future SAT exam or math homework.
1
1
92
Ralph Fischer retweeted
Dear god it's inventing German from first principles
GPT-6.1-Sol reasoning traces totally legible, nothing to worry about
162
999
50
17,330
457,722
Ralph Fischer retweeted
⚠️"Il faut que tout le monde voie ces images". Guillaume Guérin, maire de Limoges, nous fait une visite de son Hôtel de Ville sous les décombres. Je me refuse désormais à nommer ces gens-là d'émeutiers ou de casseurs. Ce sont des terroristes.
1,199
20,228
791
51,158
1,047,236
Ralph Fischer retweeted
Just watched the @netflix Instadocs about the @OpenAI cyberattack on @huggingface. Super disappointing, as they went for maximum anthropomorphism of the technology and full doomer mode. I spent more than two hours with them trying to correct all their mistakes and ground this in reality, explaining that agents are technological systems, not aliens. But it looks like the drive to scare people for maximum views was too strong. Also, they didn’t mention the lack of containment and monitoring of the agents, how closed-model safeguards blocked our defensive work, or how we defended ourselves with AI using an open model. And that we still lack a lot of information and transparency, including full agent traces and that companies need to remain accountable for what their agents do. It ends up being a fun sci-fi movie though!
249
342
99
2,321
194,475
Ralph Fischer retweeted
Ich arbeite in einer deutschen Verwaltung und bin dort für den Bereich KI zuständig, soweit man das so nennen darf. Die Menge an Anweisungen und Regelungen, um überhaupt den kostenlosen ChatGPT-Account nutzen zu dürfen, ist so derbe, dass kein sinnvoller Einsatz von KI möglich ist außer „Schreib mir die E-Mail“. Ich höre ständig nur „Geht nicht“, „Dürfen wir nicht“, „Ich sehe nicht, wie man KI da nutzen kann“. Jetzt werden wieder Millionen an Steuergeldern rausgehauen, ohne dass wir eine ordentliche Richtlinie haben und uns dem Thema KI so öffnen, dass Bots und Agenten überhaupt möglich wären. Das Mindset fehlt für KI. Niemand hat einen Plan, wie KI eingesetzt werden kann. Es gibt Tausende Start-ups, die täglich anrufen und ihre 1000. Chatbot-Lösung andrehen wollen. Solange keine offenere Nutzung von KI möglich ist, brauchen wir nichts einzukaufen. 🤦🏻 Das Einzige, was der Staat kann, ist sinnlos Geld verbrennen, und die Firmen quetschen ihn sinngemäß aus – die Kuh wird gemolken, solange es geht. @JensenHuang hat recht, solange man jede Idee und Innovative Einsatzmöglichkeit durch Regulierungen im Keim erstickt bringen einen diese Millionen oder vielleicht sogar Milliarden Investitionen nichts. Aber letztendlich klopft sich der Staat selbst auf die Schulter, weil er hat ja was getan für KI, die Firmen schauen schon wie sie daraus Profit generieren können und die Zeche zahlt der Steuerzahler, ohne das es für ihn in absehbarer Zeit irgendwelche Verbesserungen geben wird, außer die nächste Steuererhöhung, weil irgendwo muss es ja herkommen. #deutschland #ki
JENSEN HUANG ZERLEGT MERZ Der Kanzler fragt den Nvidia-Chef auf offener Bühne, wie Europa KI regulieren soll. "Man kann nichts regulieren, wenn es die Technologie gar nicht gibt.“ Die Technologiebranche muss die Technik für Sicherheit selbst erfinden. Sicherheit besteht nicht aus Hoffnungen und Wünschen, Sicherheit ist Technik und Ingenieursarbeit. Zweitens brauchen die Unternehmen verbindliche Standards. Und erst zuletzt kommen Vorgaben von Industrie und Staat. Das Internet zu regulieren ist schwer, Internetanwendungen zu regulieren ist es nicht. Genau dort sollten wir ansetzen, bei den Anwendungen. Es gibt nichts zu regulieren, wenn die Branche nicht gedeiht. Man will doch, dass die deutsche KI-Industrie floriert, und zwar sicher und verantwortungsvoll. Wir wissen, dass wir Technik entwickeln und einsetzen können, die KI sicher macht. Wir wissen, dass wir uns als Unternehmen und als Branche an gute Standards halten können. Und wenn es dann noch nötig ist, regulieren wir." Merz stimmte zu und verwies darauf, dass Maschinen bereits aus dem AI Act herausgenommen wurden. Ein europäischer oder globaler Rahmen sei noch weit entfernt.
120
192
16
1,612
117,905
Ralph Fischer retweeted
If we want to continue progress in science, we must be prepared to acknowledge that the correct description of nature will eventually simply be too complicated for humans to understand. I do not think it is wise to make human understanding a requirement. AI is the way forward.
1,207
561
434
5,824
471,636
Ralph Fischer retweeted
🇫🇷 FLASH - Des dizaines de petits groupes pillent les boutiques les unes après les autres à Châtelet, à Paris. "C’est open-bar", s’indigne une source policière qui dénonce un manque d’effectifs. Les pillages se poursuivent dans plusieurs quartiers de la capitale. (Europe 1)
98
921
98
9,159
454,011
Ralph Fischer retweeted
Sean Carroll (@seanmcarroll) doesn’t care whether a new theory comes from an actor or a physics PhD. He cares whether its author understands the theory it would replace, the Standard Model. His advice: open by saying why physicists will be skeptical, then show why those reasons don’t apply. Full episode: youtu.be/9AoRxtYZrZo
47
20
12
236
30,988
Ralph Fischer retweeted
1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness. New preprint: biorxiv.org/content/10.64898… (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)
389
1,695
534
11,329
2,193,438
Ralph Fischer retweeted
Look at these gorgeous, dynamic, orange mitochondria surrounding the blue nucleus. A view of life from beneath a cell monolayer in a parti dish. Whatever angle you look at it, life is always dynamic and pretty special.
43
325
40
1,465
45,691
WTF @GoogleAI @GoogleDeepMind After months of patient waiting (and paying for Pro), Gemini 4 Argon finally seems usable again, yet the limit is hit in a chat with 8 simple follow-up prompts without extended reasoning. Then the model shifts to 3.5 Lite - the flow is gone, as my patience reached its limit.
1
73
Pro Plan, the model is unavailable in antigravity, and @Gemini_Notebook suddenly shares usage with the Gemini app chat. 30min of work → 4.5h waiting for the next reset. What is the 5h rhythm (on top of weekly limits) anyway? Why not 4 or 6? Or even 24, so that it aligns with the flow and adapts to the day rather than forcing the user to adapt to the reset circles? Not to mention all the bugs in the app/harness or missed integration opportunities. At least Markdown is now supported.
70