The good future, according to Jakub Pachocki
What Pachocki says a good outcome with AI looks like, drawn from the 10 recorded appearances we could find and transcribe: podcasts, one talk, an OpenAI livestream and two Polish YouTube videos. He joined OpenAI in 2017 and was not one of its founders. The page also covers whether he ever talks about Iain M. Banks's Culture, or about humans ending up as AI's pets. He is a reluctant public speaker, and his vision is narrow and consistent: automate research and the rest follows.
The short answer
Banks, the Culture and benevolent "Minds" appear in none of the 10 appearances, and no host raises them. The nearest he comes is to reject the old picture of AGI as a single all-solving oracle. He says that in 2017 it was "more of an emotional thing of this thing far up in the distance where… It solves all the problems," and that "now it's becoming clear that it's really a sequence of milestones, a sequence of new capabilities that AI acquires." Before AGI · Jul 2025 · 0:34:44 ↗transcript
No host ever puts the pets idea to him, and he never uses it. His nearest answer came when Jacob Effron, a new father, asked how the next generation should be raised: "A task for all of us is to build the AI, build the world in a way where at the end of the day, humans have the agency. Humans set the direction." Unsupervised Learning · Apr 2026 · 0:56:08 ↗transcript His worry runs the other way from the pets picture. He fears not that AI will keep all of us, but that a very small group of humans will control it.
Close to the question, but not quite
He doesn't talk about being cared for or kept by AI. But he does answer the questions around it: what is left for humans to do, who holds the power, and whether people can still oversee systems smarter than them.
"Maybe a lot of the technical challenges that we cherish right now will become more of a… past time… in that world, like people can end up with, you know, more things to do and definitely more exciting things to do."
The continuation of his "humans set the direction" answer. Today's technical work may become a pastime rather than a necessity, and the challenge shifts to "figuring out like what are the things that are important." It is the nearest he gets to humans retiring into leisure, but he keeps people in charge. Unsupervised Learning · Apr 2026 · 0:56:08 ↗transcript
"I think the high level goal setting, right? What pursuits we're chasing, that is something that will remain human. And I think that is something that a lot of people will derive meaning from."
Sam Altman asked him directly what meaning and the jobs of the future will look like. He began: "I think this is a quite philosophical question. I think it can go in many directions." OpenAI livestream · Oct 2025 · 0:57:38 ↗transcript
"I definitely expect that desired skill set to shift quite a bit over time towards this more general vision setting."
On what skills people will need to supervise coding agents. Unsupervised Learning · Apr 2026 · 0:23:56 ↗transcript
"At some point, these models will be smarter than us on practically every axis. And it doesn't mean that we will understand how they work so well. And it doesn't mean that it will be in 50 years, it may be very soon."
Machine-translated from Polish. This Is It (Maciej Kawecki) · Sep 2025 · 0:27:50 ↗transcript
His positive vision, in four parts
1. Automated research is the whole point
This is the core of everything he says in public. For Pachocki, AI's most important consequence is not chat or companionship but a machine that discovers new knowledge. He returns to it in every substantive appearance.
"The big goal that we are working towards is automating research, automating discovery of new insights and development of new technologies. I believe this is… This sort of technological, this acceleration of technological progress is going to be the most important consequence of AI progress."
The opening of his solo talk. Berkeley RDI talk · Aug 2025 · 0:00:04 ↗transcript
"The most forward to, I think, AI will be able to tell us new things about the world. It will be able to accelerate development of new technologies, of finding new cures… Automating the discovery of new knowledge will be, that is something I truly look forward to."
His answer when asked what he looks forward to most. Before AGI · Jul 2025 · 0:50:07 ↗transcript
"The kind of thing that maybe emotionally is closest to what I originally thought about as AGI is a system capable of doing automated research, of actually discovering new knowledge about… About the world, about problems that we care about."
"I think we tend to associate new ideas, fundamental technological progress with just human ingenuity… It is possible to automate. Most of this process, it is possible to have a big computer that is coming up with ideas that fundamentally change our understanding of the world. And I actually think that is not that far away."
"My picture of how that would actually look like is imagine a company of very capable researchers and engineers that is largely automated."
His answer to what AGI would mean for an ordinary user. He adds that it "can radically accelerate the pace of technical progress." OpenAI Podcast · Aug 2025 · 0:30:45 ↗transcript
"We will be able to take some kind of graphic card cluster, run a number of copy-text models on it and we will simply have a research laboratory. They will just think for themselves and produce scientific results."
Machine-translated from Polish ("copy-text" is presumably "copies of"). He goes on to say this "can start to accelerate at a pace to which we are not used to," including in medicine. This Is It (Maciej Kawecki) · Sep 2025 · 0:13:46 ↗transcript
"But we do anticipate that eventually AI systems will be capable of creating valuable artifacts of… Actually pushing scientific progress forward as we were discussing. And I believe that will be the real lasting legacy of AI."
Asked whether ChatGPT is OpenAI's product or a precursor to something bigger. He notes that OpenAI "haven't set out to build… A chatbot originally." OpenAI livestream · Oct 2025 · 0:49:47 ↗transcript
2. Dated goals: an AI research intern, then an AI researcher
In the October 2025 livestream he presented OpenAI's internal research targets, the most specific predictions on this page.
"We believe that it is possible that deep learning systems are less than a decade away from super intelligence. So systems that are smarter than all of us on a large number of critical axes."
"We are making plans around getting to quite capable AI research interns that can meaningfully accelerate our researchers by expanding a significant amount of compute by September of next year… a meaningful, fully automated AI researcher by March of 2028."
His own presentation of the targets. Sam Altman then repeated them, saying "we may be totally wrong." OpenAI livestream · Oct 2025 · 0:07:40 ↗transcript His latest transcribed appearance, in April 2026, predates the September 2026 deadline. There he says "I think we're not very far from models that can work autonomously for a couple of days." Unsupervised Learning · Apr 2026 · 0:23:02 ↗transcript
"Compute how much time you would like to spend on problems that really matter, such as scientific breakthroughs. You should be OK using entire data centers."
OpenAI livestream · Oct 2025 · 0:06:40 ↗transcript The flip side of the vision: point vast compute at a few important problems. Elsewhere he says the measure of progress is using "vastly more compute than a user would normally be willing to buy for themselves to produce technology artifacts that are useful to a lot of people." OpenAI Podcast · Aug 2025 · 0:21:17 ↗transcript
3. Wonder, cures and an "extremely happy world"
He rarely describes everyday life after AI. When he does reach for something beyond science, it is a sense of wonder, medicine, and a cautious "happy world."
"For me personally, it was watching the Dota 2 bot play endlessly entertaining games, you know, like this sort of infinite TV with new strategies coming up."
He worked on OpenAI Five. He compares it to AlphaGo's move that "feels absolutely magical." Berkeley RDI talk · Aug 2025 · 0:10:19 ↗transcript
"And if you think about this world where new science just kind of falls out of GPUs, it's a very different, a very weird place to think about. And I think that's worth reflecting on."
How he chose to end the talk, leaving the picture open. Berkeley RDI talk · Aug 2025 · 0:11:18 ↗transcript
"I think in particular we see pretty incredible results on medicine, which is very encouraging. I have high hopes about that."
"I think also just the ability to understand so much more about the world, the wide, the incredible variety of new knowledge and new also entertainment, but also… Just intelligence that will be in the world, I think will provide quite a lot of meaning and fulfillment for people."
The rest of his answer to Altman's question about meaning. OpenAI livestream · Oct 2025 · 0:58:23 ↗transcript
"I think my general belief that there is a research path here that actually gets us to an extremely happy world has increased quite a lot."
He is talking about alignment research becoming concrete and tractable. In the next breath: "My timelines to very capable models have definitely decreased a lot." Unsupervised Learning · Apr 2026 · 0:46:34 ↗transcript
"We will get to a world where it's just very natural to be collaborating with… AI scientists that are working hard on a problem."
He argues for humans in the loop rather than a choice between full automation and mere tools. Unsupervised Learning · Apr 2026 · 0:35:55 ↗transcript
4. A good outcome needs governance, and a willingness to slow down
His positive vision comes bundled with one recurring worry: an automated research company could be run by a handful of people. He raised it unprompted in 2025 and again in 2026.
"Maybe I would picture it as a company of very skilled researchers and engineers that is essentially entirely automated. It just lives on GPUs and it can be administered by a small group of people. It can assure incredible progress, but it also bestows incredible power and responsibility… To whoever controls it."
His answer to what he looks forward to least. It is the same image as his hope. Before AGI · Jul 2025 · 0:50:36 ↗transcript
"An automated system that can be operated by a small number of people and is capable of fundamentally developing new technologies. I think just has very profound consequences for governance and just balance of power."
He calls this more pressing in the short term than aligning such systems. His lesson from the November 2023 board crisis was "that governance really matters." Before AGI · Jul 2025 · 0:37:18 ↗transcript
"What are these organizations that are so powerful and yet maybe made of only a couple of people? How to think about these things, I think, is a new question we have to grapple with as a society."
Asked what society is underthinking. He also names "the question of jobs and concentration of wealth, and I suspect this requires real policymaker involvement." Unsupervised Learning · Apr 2026 · 0:54:40 ↗transcript
"But I do think we have to be, as an industry, really prepared to take trade-offs. Possibly slow down development, depending on what we see."
This follows directly from his "extremely happy world" remark. Unsupervised Learning · Apr 2026 · 0:47:03 ↗transcript
"What is really the thing that the AI fundamentally cares about? Can it adhere to some high-level principles? What will it do if it's given unclear and conflicting objectives? Does it love humanity?"
His description of value alignment, which he calls "the most important long-term safety question for superintelligence." OpenAI livestream · Oct 2025 · 0:08:39 ↗transcript
How it changed over time
- 2017–24 Nothing relevant. The 2017 recording is a technical mathematics talk, and he can't be reliably picked out among the roughly 20 unlabelled speakers in OpenAI's 2024 o1 team video.
- Early 2025 In a Polish interview (posted September 2025, but it discusses o1 and DeepSeek R1 as recent, so it was probably recorded early in the year) he says superintelligence "may be very soon," worries about supervising smarter models, and gives his starkest risk statement: "Since the discovery of atomic bombs, the world has not yet been in such an unstable place as such a revolution could be." (Machine-translated.) This Is It (Maciej Kawecki) · Sep 2025 · 0:35:54 ↗transcript
- Mid-2025 The automated researcher becomes his central image, with concentration of power as the matching fear. He drops the idea of AGI as a single all-solving moment in favour of "a sequence of milestones."
- Oct 2025 He gives dates: superintelligence possibly "less than a decade away," an AI research intern by September 2026, an automated AI researcher by March 2028. He also addresses jobs for the first time: "think about automation of a lot of jobs as something that… What will be happening over the next years? And I think this is a big thing for us to collectively think about." OpenAI livestream · Oct 2025 · 0:56:39 ↗transcript
- Apr 2026 More optimistic about alignment ("an extremely happy world") but with shorter timelines, so he says industry must be ready to slow down. He gives his clearest statement of human agency and calls for policymakers to deal with jobs, wealth concentration and the governance of tiny, powerful organisations.
The hedges
Pachocki rarely describes a good future without qualifying it, and he avoids utopian imagery altogether:
- "Getting to a point where so much intellectual work can be automated, I think, comes with pretty big problems that I don't think have obvious solutions." Unsupervised Learning · Apr 2026 · 0:54:40 ↗transcript
- "And in the long term, it should be a good thing. But I think it can happen quite quickly." This is about intellectual work suddenly becoming cheap. Unsupervised Learning · Apr 2026 · 0:55:08 ↗transcript
- "Yeah, I don't think it's a complete solution to AI safety and alignment by a long shot." This is about chain-of-thought monitoring, the safety tool he is most associated with. Unsupervised Learning · Apr 2026 · 0:43:00 ↗transcript
- "Of course, if we are in such a place, we must have very good ways to supervise such models, which is not trivial, because they will be much better at thinking about these problems than we are." (Machine-translated.) This Is It (Maciej Kawecki) · Sep 2025 · 0:31:24 ↗transcript
- "I think that is something that is quite… Easy to get wrong in a few ways. And it's something I think we'll collectively as a society have to figure out." This is about power over an automated research company. Before AGI · Jul 2025 · 0:51:01 ↗transcript
The science fiction he actually cites
None. He names no science fiction in any of these appearances. The one book he mentions is non-fiction: Paul Graham's Hackers & Painters, which his father gave him in Polish when he was 15 ("a book by some author I didn't know called Hackerson Painters," as the transcript renders it). OpenAI Podcast · Aug 2025 · 0:36:28 ↗transcript
Some references in these recordings come from other people. Iron Man is Szymon Sidor's (on both Before AGI and the OpenAI Podcast). The Martian is host Andrew Mayne's. The "42" joke in the o1 team video is an unidentified team member's. In the October 2025 livestream, the description of early AGI as "this oracular thing in the sky" and the imagined data centres curing cancer and colonising space belong to Sam Altman's part of the presentation, not Pachocki's.