2026-09-30 · Kai Lentit · 03:49
“A pause now would lock in their lead. That would be unsafe.”He opposes pausing AI development while rivals are ahead.
Dario Amodei on ai progress
What Dario Amodei said, newest first. Changes are found by AI and checked before they show; every quote opens at its moment.
2026-09-30 · Kai Lentit · 06:01
“We're building a system to stop models like that.”He says Anthropic is building a system to stop models with no off switch.
2026-09-23 · Reuters · 07:45
“The trajectory only needs to continue for a tiny period longer, one or two years, maybe less, to reach what I've called a country of geniuses in a data center.”He expects AI to reach extremely powerful, human-level-research-like capability within a couple of years or less.
2026-09-23 · Reuters · 10:20
“I called for a collaborative effort to pace the frontier, not to stop AI progress, but to work together to make sure the industry and the world can proceed at a rate where we can be sure that safety keeps up with capability for everyone in the industry.”He recommends the industry and governments cooperate to pace frontier AI development so safety keeps up with capability.
2026-09-23 · Reuters · 09:59
“We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe.”He says Anthropic will slow development as much as necessary to ensure each AI technology it releases is safe.
2026-09-16 · USA TODAY · 07:19
“Even if that were the case, we're making use of maybe only five or 10% of what the possible value of the technology is.”He believes only a small fraction of AI's potential value is currently being realized.
2026-02-20 · Dwarkesh Clips · 02:35
“I simultaneously agree with you agree that it's a reason why these things don't happen instantly, but at the same time I think the the the effect is going to be very fast”He thinks the productivity effects of AI will arrive very fast, even if not instantly.
2026-02-19 · ET Digital · 01:51
“there are only a small number of years for AI models surpassing the cognitive capabilities of most humans for most things”He expects AI models to surpass most humans' cognitive capabilities within a small number of years.
2026-02-19 · ET Digital · 01:37
“AI has been on an exponential for the la for the last 10 years um and as part of a sort of Moore's law for intelligence”He holds that AI has been improving on an exponential curve for the past decade, like a Moore's law for intelligence.
2026-02-18 · Dwarkesh Clips · 00:52
“The most surprising thing has been the lack of public recognition of how close we are to the end of the exponential.”He believes the public underestimates how close AI is to the end of the exponential.
2026-02-18 · Dwarkesh Clips · 05:20
“And so we're seeing the same scaling in RL that we saw for pre-training.”He believes RL scaling follows the same pattern as pre-training scaling.
2026-02-15 · Dwarkesh Clips · 00:12
“It is hard for me to see that that there won't be trillions of dollars in revenue before 2030.”He expects trillions of dollars in AI revenue before 2030.
2026-02-15 · Dwarkesh Clips · 00:28
“in 2028 we get the the real country of geniuses in a data center.”He considers a 'country of geniuses in a data center' plausible by 2028.
2026-02-14 · Dwarkesh Clips · 02:02
“It will be faster than anything we've seen in the world but it still has its limits.”He believes the diffusion of AI's economic benefits will be faster than anything seen before but still has limits.
2026-02-13 · Dwarkesh Clips · 00:16
“I think these two these two things within the existing paradigm may just be enough to get you the countries of genesis in a data center.”He believes long context and inference-time capabilities within the existing paradigm may be enough to achieve powerful AI.
2026-02-13 · Dwarkesh Clips · 00:47
“I think there's a good chance that in the next year or two we also make we also solve that.”He expects continual learning to be solved within the next year or two.
2026-02-13 · Dwarkesh Clips · 02:16
“This is a This is an engineering and inference problem, right? If you want to serve long context, you have to like store your entire KV cache.”He thinks extending context length is an engineering and inference problem rather than a research problem.
2026-02-13 · Dwarkesh Clips · 03:03
“If you train at a small context length and then try to serve at a long context length, like maybe you get these degradations.”He believes qualitative degradation at long context may come from training at a short context length while serving at a longer one.
2026-02-13 · Dwarkesh Patel · 13:51
“On the basic hypothesis of, as you put it, within ten years we'll get to what I call a "country of geniuses in a data center", I'm at 90% on that.”He expects a country of geniuses in a data center within one to three years, and with 90% confidence within ten years.
2026-02-13 · Dwarkesh Patel · 42:24
“There's a good chance that in the next year or two, we also solve that.”He expects continual learning to be solved within the next year or two, though it may not even be necessary.
2026-02-13 · Dwarkesh Patel · 1:07:14
“It is hard for me to see that there won't be trillions of dollars in revenue before 2030.”He expects the AI industry to reach trillions of dollars in revenue before 2030.
2026-02-12 · Interesting Times · 04:29
“And so the thought is —could A.I. accelerate all of this and could we really cure cancer? Could we really cure Alzheimer’s disease?”He expects AI to dramatically accelerate biology and medicine, potentially curing major diseases.
2026-02-12 · Interesting Times · 08:21
“So I can see a world where A.I. brings the developed world GDP growth to something like percent or 15 percent”He expects AI could raise developed-world GDP growth to unprecedented levels.
2026-02-12 · Interesting Times · 53:26
“We don’t know if the models are conscious. We’re not even sure that we know what it would mean for a model to be conscious or whether a model can be conscious.”He takes a precautionary stance on possible AI consciousness.
2026-01-21 · CNBC Television · 02:21
“I ACTUALLY DON'T THINK SOME OF THESE OTHER MODELS HAVE ENOUGH INCENTIVE TO GET SMARTER, HAVE ENOUGH INCENTIVE TO BECOME CAPABLE OF MORE TASKS ACROSS THE BOARD”He believes consumer-focused AI companies lack strong incentives to make their models smarter and more capable.
2026-01-20 · DRM News · 02:22
“if I had to guess, I would guess that this goes faster than people imagine.”He expects AI progress to go faster than people imagine.
2026-01-20 · DRM News · 22:45
“I prefer Demis' timeline. I wish we had 5 to 10 years”He wishes AI development could be slowed to a five to ten year timeline.
2026-01-20 · DWS News · 01:06
“it's always hard to know exactly when something will happen but but I don't I don't think that's going to turn out to be that far off”He expects AI to reach Nobel laureate-level ability across many fields around his previously stated 2026-2027 timeline.
2026-01-20 · Bloomberg Live · 08:06
“I think there's a good chance that even happens in the next year or two. Again, it seems far away. But but this property of exponential that they like, they they they catch you by surprise. And if it's not a year or two, I think it's pretty likely that it's, you know, at least less than five years within the 2020s.”He expects AI models to become smarter than humans at almost everything within one to two years, or at least within the 2020s.
2026-01-20 · The Wall Street Journal · 06:10
“I don't think there's an awareness at all of what of what is coming here and the magnitude of it”He thinks governments and businesses are not doing enough to prepare for AI's impact.
2025-12-09 · Bloomberg Podcasts · 02:59
“We think these tools are are are are revolutionary. We've seen that from their use inside Anthropic and many enterprises already.”He believes AI tools are revolutionary and will spread across the enterprise world.
2025-12-09 · Bloomberg Podcasts · 08:38
“there's this integration layer that's missing and and you know, I think these tools are designed to facilitate that integration layer and and make it easier.”He believes standardizing how AI models connect to enterprise data is key to enterprise adoption.
2025-12-03 · The New York Times · 21:29
“I think scaling is going to get us there.”He expects scaling to deliver increasingly capable models without a distinct AGI moment.
2025-12-03 · The New York Times · 31:01
“Saying that for 10 years we won't regulate that technology, it's like it's like saying I'm driving a car, I'm going to rip out the steering wheel because I don't need to steer for 10 years.”He rejects calls for a long moratorium on AI regulation.
2025-11-18 · Microsoft · 01:43
“both of us believe in choice. And we're excited to bring our models as a choice to Microsoft Azure.”He believes offering choice of models across clouds matters.
2025-11-18 · Microsoft · 02:53
“We think this is going to be the beginning, just the beginning of a very long partnership.”He sees the NVIDIA partnership as the start of a long-term collaboration.
2025-10-20 · Anthropic · 04:28
“But in a year they might be.”He believes AI models will be capable of running processes end to end within about a year.
2025-10-20 · Anthropic · 05:02
“if you start preparing now for the large change as the models are getting better, then you know, you may save years of time.”He recommends enterprises prepare now for large AI-driven change rather than waiting for models to fully mature.
2025-10-20 · Anthropic · 04:32
“don't get too seduced by, oh, we can make these little hill climbing gains by, you know, doing this part and that part, let's start preparing to do the whole thing end to end.”He thinks incremental AI insertion into parts of existing processes is less effective than planning for full end-to-end automation.
2025-10-20 · Anthropic · 05:16
“if you do two year long projects, plan for where the AI is gonna, I mean, that sounds like an obvious thing to say, but I think it actually takes a lot of courage and foresight to do that.”He says long-term projects should be planned around where AI capabilities will be, not where they are today.
2025-10-19 · Salesforce · 16:37
“every 3 months we make a model that's smarter than the month than the model we made before. And indeed, the models are slowly getting to where they're, you know, better than most humans at most things.”He believes models are steadily becoming smarter and will eventually be better than most humans at most things.
2025-10-19 · Salesforce · 13:53
“I think it's a real thing, which is being able to do tasks end to end in a way that was not possible before.”He sees agentic, end-to-end task completion as a real and important development beyond the hype.
2025-10-19 · Salesforce · 31:58
“I just think AI is the right thing to cure them”He thinks AI is the right tool to cure tough diseases like cancer and Alzheimer's.
2025-10-19 · Salesforce · 37:04
“Some of these deals seem a little bit fishy. They seem like they're maybe double counting.”He is skeptical of some data center deals and believes real revenue and customer demand matter more than buildout announcements.
Blumify can make mistakes. Please check important info.