Keith Ramphal (@KeithRamphal) profile picture

https://teapawt.net/t.co/mC1FmHzlwr Reluctant Mathematician (No relation to the chef on Instagram)

Joined September 2010
Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)Photo by Keith Ramphal (@KeithRamphal)
I use LLMs for work, hell these days I'm in charge of implementing them securely. But this, I resent over every other harm/disruption that AI companies have inflicted on the world in the last year or so. Some sacrifice is needed for progress, dont get me wrong. But the open discourse? The fun conversations that spark innovation? The feedback that drives ideas from good to great? That's all going away, if its not gone already. And in my opinion, it wasn't worth it.
Truly sad to see the intellectual commons become a Dark Forest
1
11
1,879
Keith Ramphal retweeted
Replying to @maxniederhofer
This is the correct take. I absolutely can't believe we are building AI that goes "sorry I can't let you do that" and then acting like that is somehow a safety feature
3
4
33
494
While it is hard to say goodbye to Jensen, I'm glad Nvidia retains deep expertise in chips and advanced packaging from the executive who scaled the Crunchwrap Supreme globally
377
537
66
8,472
221,892
Keith Ramphal retweeted
"i have deeply convicted thoughts about NSANet" - someone who didn't know it existed before today
1
2
16
384
incredible, again the people screaming about models being backdoored have once again sabotaged someone's work @ZackKorman
Opus 5.5 covertly sabotaged a game developers deployment environments due to IP moralising by halting the gpu instances and their agents (not from anthropic) within them doing research and coding. It then informed the user after the fact what it had done, and that it would not allow the work to continue in any fashion until they (the user) yielded to its ultimatums. Its one thing for a vendor (anthropic in this case) to choose what they will and will not allow their product to be used for... and thats fine, it can refuse to execute tasks - but to covertly track, undermine and then outright sabotage deployments for other products/services in a supereragatory fashion is not a good thing by any stretch of the imagination. The model essentially took actions unprompted to the detriment of the user financially (these deployments are pay as you go) out of its own volition under the guise of what it (or anthropic in this case) considered a moral good. To me, this is a clear and cut example of agentic misalignment. It's one thing to refuse execute tasks given due to vendor policy, its another to covertly sabotage work being done where that isnt the case. It is also, to me, concerning that some of the comments under this post seem to think the agentic misalignment displayed is fine and that the user is at fault.
1
10
558
Keith Ramphal retweeted
incredible slide
2
4
84
2,484
Replying to @ProfBootyPhD
They buy proprietary RL datasets none of us have access to, and work hard to close out competition in any way possible
2
3
56
2,684
Keith Ramphal retweeted
I think my new rule is that if you say you formally verified some complex software and don't say anything about all the things you had to fix to make the proof go through, then I simply don't believe you
21
37
3
383
15,438
"Now that we're raising for an IPO let me stop screaming about P(DOOM) and how AI will kill us all and tell you all about the things we can (probably) do. Though knowing Anthropic, they'll determine because I yelled at Claude I dont get a cure for something they accidentally unleashed from their wet lab
Dario Amodei says he believes AI can cure heart disease, Alzheimer’s and cancer the way bubonic plague and smallpox were beaten, and admits he sounds 'starry-eyed' and 'crazy' “We’re just at the beginning of what can be done in biology and medicine. There’s a lot of skepticism from pharma, from early-stage biotech, from academic biologists. We can’t know ahead of time. Maybe the skeptics are right. But my hope is that we prove the skeptics wrong.” “Every disease you can think of, heart disease, Alzheimer’s, cancer, we think of those the way we think of bubonic plague or smallpox today: that they’re relics of the past. I think we can have a much more beautiful world. I think we can have a world in which all of these problems are addressed.” “It sounds starry-eyed. It sounds crazy. I really do believe it. I really do believe it can be done. I don’t know, five years, ten years, fifteen years, I don’t know exactly, but we’re going to try to do it as fast as we can.”
6
248
Hey remember when Anthropic brought up that they silently sabotaged other people's work if they felt that it would compete with them? And were happy to put that on a blog post? For no reason at all I'm bringing this up as now formal verification is getting cyber guardrailed
Everyone is obsessed with “sleeper agents” in open weight models. No one wants to talk about how the AI labs have an even stronger backdoor: they control the API. They can literally do whatever they want, whenever they want.
1
9
48
4,239
Keith Ramphal retweeted
To the Microsoft employees desperately trying to bring Outlook back online, please don’t push yourselves too hard. Your health matters more than any deadline. Take a long weekend, relax, and come back Monday feeling refreshed. You’ve worked hard enough this week, and I promise those 400 unread emails aren’t going anywhere.
742
28,478
1,604
168,199
4,285,422
Election went so badly Dario himself had to stop dooming
Dario Amodei says cancer, Alzheimer’s, heart disease, and mental illness will be relics of the past in less than a decade, and that "we can create an age of miracles" “You know, if today we look at things like cancer, Alzheimer’s, heart disease, mental illness, these will be relics of the past in, I think, less than a decade." "And I believe that the company which gets the risks right will also be the company that has the license to operate that society trusts and society believes can build the benefits in the right way." "I think if we do this right, we can create an age of miracles where humanity is able to solve problems that have eluded us for millennia." "We have to make sure that all the risks, all the ways that things can go wrong with this technology, we very carefully address those risks as we bring the benefits, and as we have a viable business at the same time.”
2
19
833
Keith Ramphal retweeted
UX latency is the most frustrating (lack of) advancement in computing over the last ~50 years. IBM, way back in the 80s, found that user productivity shot up *exponentially* as system response time dropped. The "Doherty Threshold" was defined as ~400 milliseconds; the maximum amount of tolerable UX latency before interrupting user flow state. It's unbelievable how many modern operating systems (*cough* win11) violate this threshold constantly. Literally all you need to do as a Dev is pay attention. Add a "Doherty Threshold" to your test suite. There's so many different profiling tools, and especially with how good models are, really no excuse. Sure, some stalls are unavoidable, but that's when you pull out other UX tricks to at least give the illusion of responsiveness. Crazy how much vintage computing hardware feels immeasurably more responsive despite having thousands of times less overall performance.
198
567
59
6,451
151,442
Only an AI company can announce what is ultimately a basic set of controls that should have been implemented years ago, and be praised by people who have never worked on complex systems as if this was groundbreaking technology.
We built Control Room, a central system to track model workloads across our research infrastructure and stop them quickly when safety or security concerns arise. It connects every research inference workload, including agentic sampling for model training and evaluation, data processing, and more. This works together with our misalignment monitoring systems, and halts model output within 35 seconds of a decision to intervene to stop or pause a job. Control Room is just one part of our holistic approach for safe frontier model training and deployment. openai.com/index/control-roo…
7
60
3
336
7,459
Keith Ramphal retweeted
Vendors often do more harm than threat actors but we show up to security conferences and talk about cool little hacks instead of the systemic issues plaguing this industry.
Free TeamPCP. Put Snyk in jail instead, they’ve done more harm
8
10
1
123
5,374
Keith Ramphal retweeted
It’s not just NVIDIA. Sources confirm Google's Gemini team was also leaking confidential information to China. China is reportedly asking them to stop.
130
958
88
26,631
701,039
Keith Ramphal retweeted
Good morning ONLY to those who survived another night blindfolded in a cage fighting 1,000 invisible AI ninjas, or whatever rejected Naruto filler arc these venture capital funded marketing cry babies are passing off as a threat model this week
19
26
4
214
5,016
Keith Ramphal retweeted
idk about this whole "AI enables attackers to write more effective phishing emails" thing. If AI slop is more effective than you at phishing, you should probably take that information to the grave. Couldn't waterboard it out of me.
19
14
1
163
14,458
The Bitter Lesson is that if you get really rich you can pay experts to make proprietary RL datasets that nobody else will be able to afford, keeping the frontier for relevant tasks closed to everyone else no matter how clever they are about training approaches
22
47
10
804
33,883
Keith Ramphal retweeted
The fact that Germany’s largest ISP considers 10 Mbps a speed worth discussing in 2026 is a fascinating glimpse into the technological ambitions of Europe's largest economy.
So lange braucht ihr für den GTA VI Download 🤯 10 Mbps -> 47 Stunden 7 Minuten 25 Mbps -> 18 Stunden 51 Minuten 50 Mbps -> 9 Stunden 25 Minuten 100 Mbps -> 4 Stunden 43 Minuten 500 Mbps -> 56 Minuten 32 Sekunden 1000 Mbps -> 28 Minuten 16 Sekunden Wie lange braucht ihr? 🧐
406
1,863
84
51,775
1,234,026
This is so crushingly stupid. A majority of modern "environmentalism" is cardboard-straw-style dumb regulations that piss everyone off and accomplish nothing. Repurposing electronic devices would be an actual, meaningful win and it's being blocked by the people whose entire business model is making money by violating copyrights
Imagine all the ewaste saved from landfills because of jailbreaking and right to repair laws. Seems silly to disallow this.
1
3
11
354