Going rogue. Breaking containment. Misalignment. P(doom). As the industry booms, an entirely new vernacular has emerged in the debates over how consequential artificial intelligence really is.

Within that new parlance, two phrases have become focal points: stochastic parrots and existential risk. And at the center of those is one prominent AI researcher who has never stood down from a fight, Timnit Gebru.

Gebru came into the public eye several years ago after sparring with Google over a research paper she coauthored that called out biases in the company’s AI, saying that LLMs basically parroted their training data and risked perpetuating biased viewpoints. The contested paper resulted in Gebru’s departure from the company, and led her to found an institute that investigates harms perpetuated by technology and supports the creation of unbiased tech tools.

She has also authored a new book, Deep Unlearning: The Rise of AI and the Radicalization of a Tech Idealist, expected to ship early next year.

More recently, Gebru has spoken out against a faction of the industry that believes AI is so powerful it could destroy humanity. In turn, some members of that community, including an Anthropic cofounder, have alleged that Gebru’s earlier research about stochastic parrots is no longer relevant, and that AI does have the ability to “think” or reason. (In an interesting twist, or you might say horseshoe, Gebru’s beliefs about the cult-like aspects of AI’s effective altruism community have now aligned her with right-wing entities who have also lambasted these EA groups.)

The AI debate is no longer just about the technology itself, but about ideological groups and strategic narratives, and Gebru believes these narratives are a “harmful distraction” from the real issues with AI. I recently spoke with Gebru about what she believes these arguments are distracting from, and also got her response to critiques that her earlier research underestimates the AI of today. This interview has been edited for length and clarity.

LAUREN GOODE: Timnit, thank you so much for being here.

TIMNIT GEBRU: Thank you for having me.

I'm gonna assume that a lot of people know this already, but for those who don’t, a few weeks ago, a young AI researcher at Anthropic stepped down, and in his resignation post on X he indicated that a lot of people within the company feel that we are facing this major existential risk with AI. A few people responded, and there was this whole pile-on and discourse about the real threats of AI.

When you and I spoke in the immediate aftermath, you said you felt that that particular narrative was a distraction from the real problems we’re facing with AI. Can you unpack that a little bit?

I would even go farther than that and say that it’s dangerous. It’s not just a distraction, but it’s harmful.

Before I answer your question, I want to give a little brief history of how this narrative has been going for a long time. Elon Musk and Peter Thiel have been saying this is going on since 2013. Prominent people have been saying that AI is an existential risk to humanity. It is exactly the same rhetoric every three years. For example, the Future of Life Institute …Wait, what is that?

The Future of Life Institute is an institute that was founded by Max Tegmark, a physicist at MIT now, and Jaan Tallinn, a billionaire who led Anthropic's Series A funding. He was a Skype cofounder, correct?

Yes. Jaan Tallinn also funds METR, which is an auditor, a so-called third party, whose report about OpenAI’s so-called rogue agents hacking Hugging Face went viral.

So to the public it might seem that there are so many different entities coming together saying the same thing. But the same billionaires who founded and funded Anthropic, the ones who stand to benefit from its IPO more than anybody else, also founded and funded the institutions warning about the, quotation marks, “existential risk” of AI. They also founded and funded the third parties that are being heavily cited right now as if they are an independent entity.

Tell us why you thought the “machine-god narrative,” as you put it, is a distraction.

It’s even more than a distraction. It’s harmful. The reason I was telling you all these facts is that these people, the funders, the founders, the investors who stand to benefit and profit the most from these companies IPO-ing are saying this, that they have been the ones seeding this narrative going back multiple decades.

So people should ask why. If I stand to get lots of money from a particular thing, why do I seem to be the person saying this particular company might build something that kills us all? It sounds counterintuitive.

This is why I really have a lot of respect for [former Federal Trade Commission chair] Lina Khan, because one of the things she said is that there’s no exception to the current laws we have for AI companies, and they talk as if there is an exception. You’re saying that whatever you’re building is so powerful that it's beyond anything we’ve seen before.

Right. It’s self-aggrandizing. It’s saying we’re the future-makers here, but the future’s gonna be really scary if we can’t continue to build this.

You have to pay attention to what they say in the news cycles. When they say that AI can cause an existential risk, they also say that AI can stop climate change, bring us world peace, and eradicate poverty.

AI can do all of these things because it’s powerful. It can bring us utopia, but also if the wrong people do it and we don’t have guardrails or whatever, it can also just kill us all. So the same people believe these things. There’s only a few people who might be in one of these camps.

When you’re telling people that you have super powerful machines, what you’re speaking to, first of all, is to your investors. You’re saying, “You wanna get your hands on these super powerful machines.”You’re talking to governments saying, “You don’t want your adversaries to get their hands on these super powerful machines. You wanna get your hands on these super powerful machines.”And you're also telling any regulating bodies that whatever regulation they’re thinking about should be about these fictional super powerful things.

If I’m talking about potentially eradicating all of humanity, things like pollution, data centers, they sound kinda small potatoes, right? You're talking about copyright protection for artists, actual lawsuits that are going on right now? The government should not waste its time thinking about this stuff. And if you waste your time regulating us on these small topics, you risk China getting its hand on the super powerful machines. And you don't want that. Trust us, we can build a machine-god that can help us, and we’re telling you we wanna be regulated. Right. That’s the first one. The second issue is you can always abdicate responsibility.

In our prior conversations, we’ve talked about the use of certain language or phrases to describe this modern era of AI and how you take issue with some of them. So, for example, in our previous conversation, you said something about autonomous weapons. After that was published, you told me you later regretted saying that because you would’ve phrased it differently. We’ve talked about P(doom). We’ve talked about how when you say that agents have gone rogue, such as in OpenAI scenarios, that you’re essentially removing culpability from the engineers who built those tools.

I wonder if there are groups that are actually more aligned than they realize around safety and transparency, but who are deploying different phrases to describe what is essentially the same thing.

One of the reasons that I regret saying these words is not that I have any issue with the term “autonomous weapons.”It’s a legal term. Even land mines are considered autonomous weapons, legally. It’s just that people in the so-called existential risk camp have co-opted it to mean something different.

When I said the term, I knew. I’m like, “Huh, I wonder if this is gonna be taken in that way.”And it was, because the people talking about existential risks of AI took it to mean what they’re saying, right? Which is like some kind of a superhuman, superintelligent machine doing stuff on its own.

I happened to be at the UN General Assembly this past week. I was part of a side panel conversation, so I was not allowed into the General Assembly nor the Security Council. But a lot of top tech leaders spoke. Both Sam Altman and Dario Amodei talked about the need for global cooperation to address the existential risks to humanity. Dario reportedly said, “If managed poorly, I even believe AI could be a risk to humanity as a whole.”

His competitor, Sam Altman, said, “We could lose control of the future to AI.”So explain how it is that these folks who have these companies that have commercial interests in selling AI are now here on the world stage at the UN saying, “I don’t know. There could be some risks here.”Is this part of regulatory capture? Are they gonna be looking for ways to self-govern before there’s an actual regulatory crackdown on this?

Absolutely, it’s part of regulatory capture. What can be done? What kind of governance do you think is needed for AI? So this regulatory capture has been happening for a long time.

In 2023, same thing. Sam Altman said the same thing. We need world cooperation, et cetera, et cetera. The EU AI Act regulated, and then he threatened to pull out of the EU. So you just have to see what they’ve actually done. When there’s regulation that actually regulates them for real things that they’re doing and holds them liable, they threaten to pull out or they lobby super hard to water down that regulation.

On the other hand, they’re going around telling these multilateral bodies that there has to be world cooperation, et cetera, et cetera. And then again, what is this doing? It is selling themselves as organizations that are creating so-called superintelligence, right? So that’s already marketing.

They’re also making everybody scared of anybody else who might be creating such things. So they don’t want open-weight models from China. They don’t want this competition. I am not following Chinese regulation that closely, but they’re not talking about existential risk.

They’re talking about deepfakes, and they’re talking about real things that need to be regulated. So the first one is marketing yourself as creating super powerful, unprecedented things for which we don’t have existing regulation, which is not true. We have existing regulations. But the second one is what they’re doing when someone tries to enforce the existing regulation, they either threaten to pull out or they lobby hard so that they don’t have this existing regulation. So the reason I keep on going back to 10 years ago, five years ago, three years ago, is that it’s the same. In my book, I say it’s “the same movie on repeat with different heroes.”So what is the solution for governance right now if you had to propose it?

Very, very simple things. First of all, I’m not a legal scholar, but former FTC commissioner Lina Khan outlined five things. One is deceptive marketing practices.

You know, there’s a law for deceptive marketing practices, and you can go after companies for that.

Two is transparency, documentation. Before you put something out there, you should be able to tell us where all the data came from and actually document it. This simple thing they don’t do.

I’m gonna tell you that they will fight tooth and nail to do the simple thing of documenting data.

Labor exploitation of data workers, that’s another one that they don’t wanna talk about. We are here in the world, in the clouds talking about superintelligence, where you have hundreds of millions of people around the world painstakingly labeling data, even pretending to be chatbots. Right. 404 Media recently reported that the new Meta Muse chatbot, when you go to use that for some use cases, actually has a person on the other end responding.

Yeah, very simple. Data transparency, labor exploitation, you should not be able to steal data from people. Even the first three things I talked about, data transparency, documentation, labor exploitation, if they had to abide by laws like that and they were not allowed to steal data and not document it—right now, the market calculation is not working, but with these additional measures, it just would not work whatsoever. You would automatically slow them down and have to make them accountable for something.

Back in 2021, you coauthored a paper that was titled “On the Dangers of Stochastic Parrots.”Google had approved it initially, then it was being reviewed. There were parts of it that were in dispute. At some point you said, “Look, if you want me to remove my name from this, I’m not gonna be a part of this.”You ended up decamping from Google. That paper was later presented at the 2021 ACM Conference on Fairness, Accountability, and Transparency, and people still reference it.

Can you briefly explain, for someone who’s never heard of stochastic parrots before, what it has to do with the AI we’re using today, what it means?

Stochastic parrots is a metaphor to help people understand what large language models do. Large language models are models that are trained on vast amounts of textual data on the internet, and they are trained to output the most likely sequences of text given their training data. They power most of the chatbots we see today, whether it’s Claude or ChatGPT.

When I wrote this paper ChatGPT hadn’t even come out yet, but we saw the race to build larger and larger language models. So it’s the danger of building larger and larger language models that we were describing in this paper.

That they would essentially parrot people? So, to parrot is to repeat back without understanding, right? So there was this whole existential risk narrative happening back then too, if you can believe it. You know, OpenAI had claimed that GPT-2, the precursor to GPT-3 that powers ChatGPT, was too dangerous and too powerful to release. There was this whole conversation about whether GPTs can be ethical or are they creative and all this stuff. So we really wanted to ground the conversation in the real issues.

One of these issues is the environmental catastrophe, which a lot of people are now seeing, and that was one of the main sections that Google people were unhappy with. The environmental cost. The other one is not documenting your data because you say you have too much data to document.

The other one is deceiving people into believing that there is a mind behind the textual outputs that they’re interacting with. That’s where we really wanted to explain that these systems are parroting the patterns of their training data.

Mm-hmm. It’s very dangerous when you’re outputting text like that because when you have plausible-sounding text or very fluent text, there’s so many different kinds of issues that can occur besides you believing that there’s a mind behind a machine. In that paper, we gave an example of this Palestinian man writing “good morning,” which was translated to “attack them.”

Because of that grammatical correctness there were no cues that the translation could be wrong, and people believed the translation. So the other issue of believing there is a mind behind the machine is what we call automation bias.

You over-trust automated systems, and if you believe that this thing is an all-knowing machine, then you’re gonna over-trust the errors that you get, right? It’s funny. I must be too much of an elder millennial because you say that there’s too much automation trust, and I’m so distrustful.

If I call the bank and it’s a robot I’m like, “Nope, nope.”“I don’t want to talk to it.”Exactly. One of Anthropic’s cofounders, Jack Clark, recently posted something on X. He basically put stochastic parrot in quotes and said it was a memetically fit cognitive virus that spread from 2021, when your paper was out, to 2025. It temporarily blinded many gifted people to the nature of AI progress, burned up crucial years of research, he says. He later says the use of this frame causes people to materially underestimate what AI systems can and can’t do. When you saw Jack’s post on X, what was your initial response to that? I was not surprised, let me just tell you.

This is a talking point that [effective altruists are] telling lawmakers now, because every time I said something, they were like, “Oh, you should not take seriously someone who still takes the stochastic parrots thing seriously in 2026.”The idea is that the research is outdated, right? Yeah. And it’s not. It’s so ludicrous that we’re even saying this, because these are definitions of what large language models are. What large language models are has not changed, will never change.

Large language models are large language models. Now, you might have large language models in a separate system. They now have reinforcement learning agents. But our paper was about large language models, and that’s never changed. And these chatbots still have large language models as a basis. It’s so ridiculous that Jack Clark is talking about overestimating systems, because what I’m seeing is the examples that I just told you. It's over-trusting these systems, not having checks and balances, and ending up misdiagnosing someone’s breast cancer to the wrong side.

There’s a direct line between LLMs being stochastic parrots, essentially, and giving a medical misdiagnosis using an AI tool because why? How does one lead to the other?

The stochastic parrots, they don’t understand what’s inside the text, so you cannot expect them to be factual. Even if you see something like the AI Overview. I was calling an oncologist friend of mine to ask about a specific medication and whether it was appropriate for a specific use, because I read the academic paper saying that it was not, and I wanted confirmation. And my oncologist friend was like, “Oh, look at the Google Overview. It says that it’s appropriate.”But it turned out not being.

I encounter that all the time with Google AI Overviews. Because they are stochastic parrots trained to give you the most likely sequences of text based on their training data.

Emily M. Bender has been trying to say this in so many languages for a long time. Also one of the coauthors on your paper, correct?

Yeah. To say that the research is outdated is ludicrous because right now, as we are hyping up superintelligence and all that, there are news stories that are going unnoticed, which are about the complete opposite scenario that is actually happening in the real world.

My understanding is that some of the biggest critiques people have had about focusing on research from that era is that it may not be incorporating the thinking or the reasoning or even the recursive intelligence …There is no reasoning. There is no recursive intelligence. Is there none? How is there none? This is something that, by the way, at the Berkeley AI conference earlier this summer I heard someone from Google talking about recursive intelligence.

Of course they are! The New York Times just did a big story about how the scientists and researchers right now, that’s what they’re looking towards, recursive intelligence, the AI learning from other AI. Is that a reality? It sounds like you’re saying that’s not a reality.

No, because there’s one problem with AI researchers, which is aspirational naming and aspirational stuff. So machine learning, that’s aspirational naming. But machine learning is real. But the naming is aspirational.

The machine is not necessarily there. So just because there’s curriculum learning in AI, it’s a field.

But let me give you a paper from my former manager, Samy Bengio, who is a superstar in machine learning, not as famous as his older brother Yoshua. I ask him, “Aren’t you tired of your whole life being, like, debunking the whole reasoning thing every single paper you write?”He quit after I got fired from Google, and he’s now head of machine learning research at Apple. If you look at almost every single paper that they have, it’s showing how if you change the benchmarks on reasoning slightly, the whole thing breaks down.

It’s not reasoning. Got it. Let me give you another example. Just because your models, the stochastic patterns that you trained to print out certain tokens, you call them chain-of-thought reasoning.

You didn’t know that they were thinking, you don't know it's a chain, you just know that these are tokens that are being printed out, but you call them chain-of-thought reasoning. Now you’re saying that they’re reasoning already.

One of the biggest crises that we have right now is actually sound scientific research. So if you look at my work, if I ever have access to the data, the code, the training data, and the evaluation data, which none of these companies give you those things—you don’t even know if they ingested that benchmark during training or not.

If you ingest a benchmark during training, it’s like studying to the test. It’s like me coming to an exam knowing what the answers to those 10 questions are already, studying that, and writing it down, right? Every time I have had access to these things, I have shown how their claims are not correct.

But then with Samy Bengio and his team’s paper, what they showed—and this is from 2025—is that you just change the benchmarks a little bit, tweak it a little bit, and it all breaks down, showing that you were just relying on the patterns. Now, some people, when I give them this example, they’re like, “Well, but we’re talking about 2026 models.”Guess what? Real evaluation in science takes time. We should not be going from press releases to lawmakers parroting those press releases and journalists repeating those claims.

If you wanna do real research and evaluation, ask for the training data, the evaluation data, and the methodology so we can all reproduce it. So the thinking, reasoning, recursive intelligence—it sounds like you still see that as something that is purely human. It’s the way our neural processes work, but the AI doesn’t work that way yet.

That’s what you believe. I don’t know if it’ll ever … Like intelligence, that’s aspirational naming. Just to take people back to these hype cycles, I sometimes play this game. I tell them a claim that was made and I say, “Is this 1954 or 2024?”

What would you say is the biggest part of your own thinking, your own research, that has evolved since the early 2020s of AI until now that has surprised you the most? I have to be actively making space for the kinds of models that I think should be built and building them, and ignore the noise. That’s the conclusion I’m getting to. That’s your biggest learning, your biggest takeaway.

Yeah. There is so much noise online right now, it’s hard to get any deep work done if you’re just literally paying attention to whatever the AI guys say.

Even if your research, at this point, is all about debunking what they’re saying, you know? It gets tiresome. It’s not fun to do that. It’s more fun to think about the future you wanna have, the technological advancements you wanna have, and work on that. What would you say gives you the most hope right now for the future of AI?

I want to tell you that we have a list called the AI Resist List, where we talk about people resisting, in the ways that they should resist: in media capture, narrative, data centers, funding, et cetera. And so we list the ways in which they’re resisting, and also people creating alternative tech futures that don’t kill our environment, that don’t exploit labor, or steal data, and instead are actually actively helping their communities, and these ideologies are spreading, right?

You see one small organization somewhere doing something, you get inspired by them, you do something different.

So, I have hope that there are a lot of people tired of what they’re seeing, and they’re actually in small circles doing something different. I believe in human agency and collective power to imagine a better future and stop bad things from happening and ban things if they are bad, right? My belief in human agency, I think, is what gives me hope. How to