together, we build an intelligent future.
Joined November 2021
- Tweets 33,882
- Following 3,881
- Followers 66,882
- Likes 9,325
4,584 Photos and videos
Ray Kurzweil says quantum computing is unlikely to deliver on its biggest promises, and we don't need it
Its outputs are still filled with errors, and despite years of progress reports, it has not broken major cryptographic systems
"AGI is coming without quantum computing"
6
4
19
993
gpt-5.5 is currently somewhere between fable and opus
so gpt-5.6 should get close to fable 5 at a much lower price, which is good enough for now
to clearly beat fable, oai probably needs a new pre-train, and i don't think anthropic will drop another step-change model like fable anytime soon
10
3
79
5,602
even if gemini 3.5 pro is only 90% as good as the latest openai and anthropic models
that would still be acceptable
because when you pay for one of google's plans, you get a much broader ecosystem of tools and integrations:
drive storage, notebookLM, img gen, and several other multimodal capabilities
4
4
62
3,226
BREAKING: private US government–anthropic conversation has been leaked
anthropic: our models are getting too powerful, so AI needs to slow down before AGI becomes a serious risk
US government: understood. we'll start by banning yours
anthropic: ðŸ˜
42
17
205
9,589
unpopular opinion
but if we want AI acceleration, we need to support open-source and people like sam altman or elon musk that are bold enough to put powerful tech in public hands
you don't have to like them
google had a chat model before openai, but only moved after chatgpt forced their hand
17
6
91
3,963
Microsoft Mustafa Suleyman says science and technology are the engines of everything we are as humans
The way to do that is training models to be great generalist reasoners that can invent new science
The test is simple:
"does this accelerate human progress? if not, the technology should be rejected"
7
14
40
3,216
Microsoft Mustafa Suleyman says science and technology are the engines of everything we are as humans
The way to do that is training models to be great generalist reasoners that can invent new science
The test is simple:
"does this accelerate human progress? if not, the technology should be rejected"
7
14
40
3,216
anthropic played a major role in gatekeeping AI access
they didn't have to push this level of fearmongering around mythos 5 when gpt-5.5 is nearly as capable
you can't fearmonger on one side and then complain when there are consequences from that same fearmongering
9
10
118
4,858
now it comes down to gpt-5.6
anthropic got what it wanted
dario and the effective altruist crowd at anthropic have been fearmongering about fable's capabilities to limit public access to advanced AI
and now they're seeing what happens when the government takes their claims seriously
26
29
440
24,827
anthropic: our model is extremely dangerous and could pose serious AGI-level risks
government: interesting. let's ban you
anthropic: wait, no. we meant the open-source guys
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
13
8
75
4,124
Jensen Huang says the internet is moving from serving billions of people to serving billions of AI agents
These agents will communicate, collaborate, and solve company problems while staying guardrailed and sandboxed
"the internet may soon be used mostly by 100 billion agents working nonstop"
15
18
84
6,208
anthropic, openai, and google keep saying they are working toward superintelligence
and recursive self-improvement (RSI) is also within the pipeline, which itself could lead to ASI
so, unless people like demis hassabis are wrong
the transition probably won't take long
12
7
94
5,108
gpt-5.5 might be the best model openai could economically serve at scale
anthropic has taken the opposite path with mythos, training a much larger model that current hardware can't serve efficiently yet
almost like their gpt-4.5 moment
except this time, the pre-training actually worked
6
7
124
10,160