Karen Hao
@_KarenHao
Followers
68K
Following
6K
Media
255
Statuses
3K
ai reporter & author. national magazine award winner. words in @theatlantic. formerly @wsj @techreview. pre-order my book EMPIRE OF AI (May 20): https://t.co/wh5mCWbuwQ.
Joined February 2016
The unnamed book EMPIRE OF AI is mine, based on 300+ interviews, 7 yrs covering AI, & my time as the first reporter who got extensive access to OpenAI. I sought OpenAI's perspective throughout. For months they said it was coming. It never did. Pre-order:
there are some books coming out about openai and me. we only participated in two—one by keach hagey focused on me, and one by ashlee vance on openai (the only author we’ve allowed behind-the-scenes and in meetings). no book will get everything right, especially when some people.
31
217
1K
A story about Danish Siddiqui. In Feb 2020, I was rehearsing my TEDx talk in India when he, a fellow speaker, sat down next to me to ask about my work. I had no idea who he was. He said simply he was also a journalist. Only later did I realize the humility in that description.
71
4K
15K
As someone who has reported on AI for 7 years and covered China tech as well, I think the biggest lesson to be drawn from DeepSeek is the huge cracks it illustrates with the current dominant paradigm of AI development. A long thread. 1/.
215
3K
13K
Nine months ago, I learned that Facebook has a Responsible AI team, and has had some form of it since Cambridge Analytica. I began to wonder: what has it been doing these past three years? Nearly four-dozen interviews later, here's what I found.
125
3K
6K
Over the past few weeks, the Facebook Papers have reaffirmed that FB has fueled the spread of hate speech & misinfo around the world. But there’s a crucial piece missing from the story. FB isn’t just amplifying misinfo. The company is also funding it.
74
3K
5K
Two years ago I discovered scholarship arguing that AI is creating a new colonial world order. The idea haunted me & pushed me to investigate. Today I begin to release my findings: a culmination of reporting across five continents. Here’s my introduction.
80
2K
5K
If you're wondering what Google was so terrified of @timnitGebru publishing, I got a copy of the paper. Here's my best summary of what it says. Thank you to @emilymbender who coauthored the work for sending it my way.
100
2K
5K
I write words for a living and yet cannot find the right ones to describe the mix of grief, fear, and loss of seeing the news and thinking: I am an Asian woman, too.
30
437
4K
My heart goes out to his wife, his kids, and his colleagues. We lost a giant today.
18
229
3K
If you’re tuning into Chinese politics for the first time, here are four resources I found useful for understanding what just happened at Party Congress, China's national convening every five years to choose a new leadership team:.
30
702
3K
It was devastating to wake up this morning & learn he was killed in Afghanistan. There isn't really a moral to this story other than that he was an incredible human. He barely knew me yet treated me with such decency. I can't imagine how many other lives he irrevocably touched.
7
216
3K
On the eve of the 2020 election, troll farms were running vast page networks on FB targeting Christian, Black, & Native Americans. An internal report tracking the situation described it as "genuinely horrifying." Some of the pages remain two year later.
62
2K
2K
At the time he was preparing to fly to Wuhan the next week to document a strange new disease outbreak. I told him he was incredibly brave. He told me it was his wife & kids who had the real courage to say goodbye to him each time he left for another life-threatening assignment.
3
243
3K
Danish was in fact a Pulitzer Prize-winning photojournalist who'd covered some of the most brutal and dangerous conflicts globally. His photos are so beautiful and grotesque they make you hurt. Compared to him, I was nothing. But never once did he make me feel small.
8
212
3K
I have an investigation coming out tomorrow over a year in the making, related to the climate impacts of AI and a tech company's hypocrisy. Stay tuned 🙏.
20
217
2K
Over the course of a week-long TEDx program, we spent several hours chatting, including sharing a cab through heavy Mumbai traffic back to the airport. He told me about his incredible love story with his wife, about his adorable kids, about his compulsion to keep doing good work.
4
147
2K
I see this argument all the time from tech people: Building gargantuan AI models may be computationally, environmentally, and financially costly. But if those models then go on to solve cancer, isn't that on balance better for the world?. NOOOO. A thread.
144
591
2K
It's weird how much a few conversations can make such a lasting impression. But I was so touched by his humility, warmth, and kinship. I followed his work on Instagram religiously and thought often about when I'd go back to India and take him up on his word.
1
142
2K
For years I’ve been interviewing data annotation workers who are the lifeblood of the AI industry. For years I’ve heard the same story: the platforms they work for wield total power, leaving them precarious & vulnerable to exploitation. A horrible example of this just happened 1/.
22
914
2K
Reporting this thoroughly convinced me that self-regulation does not, cannot work. Facebook has only ever moved on issues because or in anticipation of external regulation. If we have any hope of fixing FB's problems, we can no longer afford to wait for it to do so itself.
26
500
2K
U.S. defense experts, after reading 87 pages of translated Chinese commentary, shocked to discover Chinese scholars read English.
54
325
2K
When we parted ways at the airport, he welcomed me to visit his hometown whenever. I now had a friend there, he said, and promised to take me around.
1
115
2K
DeepSeek is the flip side of the same coin: It innovated because of, not in spite of, its constraints. And now that it has upended the assumptions of the dominant AI paradigm, we should reject its costly trade-offs and seek new ways to develop AI without so many harms. 24/.
60
274
2K
To the public, Microsoft uses its reputation as an AI & sustainability leader to tell a compelling story: AI will do wonders to help solve the climate crisis. To fossil-fuel firms, Microsoft has a different message: AI will help them drill, baby, drill. 1/
17
840
2K
Black Panther makes the celebration of women and people of color look so effortless that it makes you wonder what took Hollywood so goddamn long. #WakandaForever.
10
426
2K
A @techreview investigation has found that Facebook & Google are paying millions of ad dollars to bankroll clickbait actors, fueling the deterioration of information ecosystems around the world. Many of these actors *would not exist* without these payments from both platforms.
13
681
2K
IT'S HERE!!! The biggest story I’ve ever worked on. @techreview's very first interactive ever, which walks through a concrete example of AI bias, and why it’s so much more complicated than initially meets the eye. 1/.
41
802
2K
A big question looms over generative AI: what really is its impact on the environment? I spent months investigating a single campus of Microsoft data centers in the Arizona desert - designated in part for OpenAI - in an attempt to find out. Thread.
31
672
2K
Hello!! Today is my first day at the Wall Street Journal. I’ll be covering China tech & society and moving to Hong Kong in May. So excited to work with such talented colleagues. 😍.
105
30
2K
Wait a minute. You mean to say that we don’t need to blanket the earth with data centers and coal & gas plants to maybe arrive at a future where we can wave a magical AGI wand to make all of the consequences of that go away?. Yes. This is a false trade off. Let that sink in. 16/.
6
349
2K
This work was not led by NIST. It was led by @jovialjoy, @rajiinio, and @timnitGebru. Really disappointed to see this @60Minutes segment. Joy says she spoke to the producers for many hours and yet they still chose not to feature her work.
According to a study conducted by a team of computer scientists at the National Institute of Standards and Technology, many facial recognition algorithms they tested had a harder time making out differences in Black, Asian and female faces.
16
481
1K
Oh hey my promotion is official today! I'm thrilled to say I have progressed from senior reporter to senior editor for AI at @techreview, and the best part is this is not an April's Fools joke. 🥳.
99
24
1K
This segment misses a huge detail: Bytedance has distinct apps for the China & US markets because the Chinese and US governments have distinct approaches to digital platform regulation. In other words, when you don't regulate platforms, it can harm kids. Shocking, I know.
“It’s almost like [Chinese company Bytedance] recognize[s] that technology’s influencing kids’ development, and they make their domestic version a spinach TikTok, while they ship the opium version to the rest of the world,” says Tristan Harris.
27
320
1K
After spending a few years cutting through Big Tech's B.S. about AI ethics, I've created a glossary to help you decode what all of their favorite terms actually mean. I had way too much fun working on this! Threading some of my favorites below.
22
440
1K
Much of the coverage has been focused on US-China tech competition. That misses a bigger story: DeepSeek has demonstrated that scaling up AI models relentlessly, a paradigm OpenAI introduced & champions, is not the only, and far from the best, way to develop AI. 3/.
12
185
1K
This piece was so incredibly painful to write. It's not about corrupt people do corrupt things. That would be simple. It's about good people genuinely trying to do the right thing. But they're trapped in a rotten system, trying their best to push the status quo that won't budge.
17
210
1K
The @WSJ raised a lot of eyebrows with its earlier prediction that Li Qiang, Shanghai's top party official who oversaw its months-long covid lockdown, would make the Standing Committee. We were right. Incredible scoop from the team.
Xi Jinping, Li Qiang, Zhao Leji, Wang Huning, Cai Qi, Ding Xuexiang, Li Xi.
25
194
1K
So it doesn’t matter if you're a company in the US, China, or elsewhere. DeepSeek should be a cue to pivot HARD toward investing in far more efficient methods of AI development. Even if you care nothing about community & climate impacts, it’s just better business. 19/.
5
258
1K
The Shanghai police data heist grows more insane: Experts say the database of nearly 1b Chinese citizens was not hacked—it simply had no password, allowing the thief to waltz in, wipe the data & leave a ransom note: "contact_for_your_data…recovery10btc."
22
543
1K
Facebook communications sent me a long list of so-called "factual errors" after the piece. @glichfield & I sent them a point by point rebuttal showing that they were in fact not errors. They acquiesced. If you have more "factual errors" you would like to send to me, please do.
@_KarenHao Yes, that's why your piece was a surprise to me and to many of my colleagues. From my point of view, you piece is full of factual errors and incorrect assumptions of ill intent. What happened to you?.
18
217
1K
…they take up drinking water - yes, drinking water, because of the water quality required to cool these facilities - which threatens to hike water prices for families… 8/.
7
258
1K
Thus far OpenAI & its peer scaling labs have sought to convince the public & policymakers that scaling is the best way to reach so-called AGI. This has always been more of an argument based in business than in science. 4/.
4
161
1K
First, what is DeepSeek? A Chinese firm that was able to produce an open-source AI model with roughly 1/50th of the resources of state-of-the-art models yet still beat OpenAI’s o1 on several benchmarks. 2/.
6
119
1K
Many journalists have written extensively about those externalities. When these massive data centers come to town, they distort power supplies (per @business’s @Leonardonclt, @naurtorious, @andretartar)… 7/.
4
266
1K
The fact that most Chinese scholars can consume ideas & scholarship in English but most American scholars can’t do the same in Chinese will over time become the U.S.’s greatest disadvantage.
37
143
1K
The problem is there are myriad huge negative externalities of taking this approach - not least of which is that you need to keep building massive data centers, which require the consumption of extraordinary amounts of resources. 6/.
2
133
1K
When I worked at an SV startup, we used to call an investment from SoftBank the kiss of death. SoftBank would pump so much cash into startups that it would completely kill that startup’s need to innovate or develop a financially sustainable business. 21/.
5
142
1K
I am beyond the moon to announce that today is my first day as a contributing writer for @TheAtlantic, focused on AI. I’ve missed this beat so & can’t believe I get to write about it for one of favorite magazines in the world, among such brilliant journalists. Let’s get to work!!.
76
39
1K
Now on top of the negative externalities on people’s power, water, air, and global climate, we have another: the ceding of more & more control over critical energy & water infrastructure to Silicon Valley. I wrote about this last week in @TheAtlantic. 14/
3
233
1K
The hardest and most complex story I've ever worked on is publishing tomorrow. Nine months of reporting and writing. So many interviews. So many drafts. I'm honestly in shock that it's almost over? And incredibly anxious for you all to read it. It's about Facebook.
24
55
1K
We decided to publish the report in full. The numbers are astounding. Collectively, the troll-farm pages reached 140M Americans and combined to form the largest Christian American and Black American pages by a massive margin.
10
305
858
I *strongly* disagree with this line of logic. Concentrating resources away from the marginalized in society to the richest in society so that the rich can then decide who to bestow those resources to doesn't make the world a more equitable place.
32
147
966
As I said earlier, scaling was always more about business than science. Scientifically, there’s no law of physics that says AI advancements must come from scaling rather than approaches using the same or fewer resources. Scaling is just an incredibly easy-to-follow formula. 17/.
5
169
1K
I downloaded every single AI paper abstract available on one of the largest open research databases to understand where the field has been & where it's going next. After months of data wrangling & head-scratching, I'm excited to finally launch this story.
34
420
976
We’re basically seeing this again. OpenAI with Sam Altman is so good at raising capital (incl. now from Softbank) that it's covering up the company’s technical & commercial weaknesses. See also this excellent @AINowInstitute report from @bcmerchant. 23/
4
175
1K
These pages had also achieved their reach not through user choice but primarily as a result of Facebook’s own platform design and engagement-hungry algorithm. 75% of the users exposed to their content had never followed any of the pages.
8
279
823
But DeepSeek shows - in the same moment as the Stargate announcement - that the trade off that OpenAI & co frame as wholly necessary is actually not. 15/.
1
157
991
DeepSeek now also undermines the business case for scaling. OpenAI has been burning through staggering sums of cash to keep up its scaling paradigm and has yet to figure out how to balance its checkbooks - and it turns out it didn’t need to spend so much cash. 18/.
5
158
972
…they extend the lives of gas & coal plants, worsening air quality (per @washingtonpost’s @evanhalper) - not just because fossil fuels are the fastest way to surge energy supply but because 24/7 facilities can't run solely on intermittent renewables. 9/
2
174
965
After a long two weeks since being forced out by Google, I sat down with @timnitGebru to reflect on her time at the company, the future of AI ethics research, & how to hold tech companies accountable. Thank you to Timnit for being open and vulnerable. ❤️.
9
305
869
They could mask all of the shakiness in their foundations with the hilariously large infusions of cash for quite a while — but eventually everything would topple. Necessity is the mother of innovation. Without necessity, the existential drive to innovate disappears. 22/.
6
108
920
OpenAI marshaled a global pipeline of human labor for 2+ years to enable ChatGPT & GPT-4 to exist. Workers who dealt with the worst parts are still suffering PTSD more than a year later. My story with @dseetharaman out today on the front page of @WSJ.
12
392
823
When a company like Google or Microsoft or OpenAI builds these gargantuan models so that *they,* our benevolent tech dictators, get to exclusively decide what kinds of problems they want their models to tackle, that does NOT make the world a more equitable place!.
22
117
798
OpenAI & co have come up with a funny argument for why to accept these externalities: Yes unfortunately there are trade-offs in the short term, they say, but they're necessary to reach so-called AGI; AGI will then help us solve all these problems! So trade offs = worth it. 11/.
5
109
878
And if a company does not make this pivot, that should be a major red flag for its capacity to innovate. And I don’t mean product innovation but innovation on how to develop AI models. The base of the stack. 20/.
3
95
873
I reviewed @katecrawford's new book Atlas of AI and interviewed her about why it was so important to her to write it. Tl;dr: It's a masterpiece, and I haven't been able to stop thinking about it.
13
226
782
The pandemic gave AI a chance to prove itself. But after a global effort that led to the development of ~ hundreds ~ of tools to catch covid, not a single one of them helped. Read @strwbilly's important story.
37
376
783
Rather than making fun of the lawyer, we should really be asking how these technologies are being marketed in a way that misleads people about the extent of their capabilities and use this as a case study to understand how that can cause real-world harm.
A lawyer used ChatGPT to do "legal research" and cited a number of nonexistent cases in a filing, and is now in a lot of trouble with the judge 🤣
52
241
740
There is empirical evidence that scaling AI models can lead to better performance. For businesses, such an approach lends itself to predictable quarterly planning cycles and offers a clear path for beating competition: Amass more chips. 5/.
2
82
838
I don’t just mean in foreign policy and defense. I also see this effect in technology. When I talk to American researchers, they have zero idea what’s happening in China. When I talk to Chinese researchers, they are just as fluent in U.S. research as any American.
30
75
784
We know this because of the in-depth field work of disinformation experts like @riovictoire, who has made her way inside the private forums where thousands of Cambodian clickbait actors trade tools and tips for how to get bigger payouts from FB and Google.
1
133
656
The report also includes some real nuggets. Like these charts that show how content FB knows to be violating (fuss_recommendation=R) is more likely to receive a higher rate of clicks and shares than content FB believes to be good (=G).
3
192
629
.@eileenguo got a hold of the Stanford vaccine-distribution algorithm that left out its residents & we looked at what went wrong. Of note: it's a rule-based algorithm, not a machine-learning one. And it seems like Stanford literally just forgot to test it?
15
224
707
This line of thinking has been so effective in capturing the mindshare of powerful decision makers that president Trump just last week announced the Stargate Project, a $500B private investment into data centers & other computing infrastructure for OpenAI. 12/.
4
90
749
But of course, FB continues to rank content based on engagement, including clicks and shares, which means content more likely to be violating pushes to more users as well as higher up their news feeds.
12
145
602
World, meet Alex, Bill, and Mophat, three workers whose labor was essential to filtering violence and abuse out of ChatGPT. For the first time they’re ready to tell you who they are—and how the work unraveled their lives and their families.
7
322
668
Meta hired actors during the Hollywood strike to train AI with their expressions & movements, paying as little as $300 to use the data “in perpetuity.” Nuanced story from @eileenguo illustrating how the AI industry profits from precarious labor conditions.
8
351
661
I’m so fiercely proud of being Asian-American. And all I can think of right now is, please world, don’t take that away from me. Please don’t make me become afraid or ashamed of my identity. It took me so long to reclaim it.
10
71
627
I wrote about a topic I’ve been itching to address for some time: how AI PR hype, coupled with increasingly flashy AI-generated outputs, is creating a dangerous gap between the technology’s perception & reality. A collab with @MilesKruppa.
15
185
647
My biggest takeaway from attending my first AI conference in person since ChatGPT is the level of FOMO that researchers feel about it. So many talks referencing or cracking jokes about ChatGPT. So many researchers trying to intersect their work with ChatGPT to stay relevant.
18
94
646
Holy shit.
@asmenext ASME Awards 2022: Congratulations @_KarenHao, Senior AI Editor at @techreview, recipient of the 2022 @ASMENEXT Award for Journalists Under 30 #ASMEawards.
86
14
633
On top of that name one technology in history that has successfully been redistributed completely equitably from the bastions of privilege and power to the have-nots in society. We've yet to achieve this with even the most basic resources like water, paper, electricity, internet.
93
70
627
Should Stargate follow through with $500B, it'll be the largest infrastructure spend in history and significantly accelerate the already startling pace of data center development post-ChatGPT. In turn that will significantly accelerate the aforementioned consequences. 13/.
7
86
692
I have a big story coming out tomorrow, and I’m very nervous. It took six months of reporting and writing and is based on nearly three dozen interviews. Also the longest piece I’ve ever written with close to 6,000 words. I hope you all find it worthwhile to read. 🙏.
38
16
636
Some news years in the making: My book EMPIRE OF AI, out May 20, is ready for pre-order at It tells the inside story of OpenAI as a lens for understanding the moment we’re in: the tech elite's extraordinary seizure of power and its threat to democracy. 1/
22
247
676
I have a big piece about Facebook coming out tomorrow—not part of the FB Papers—that I hope will fill a critical gap in our understanding of the company's impact on the world. So thankful to all of my sources and collaborators on this project. Hope you all read it.
5
40
589
"This is Princess Bubblegum from adventure time. She is a leader and a scientist." -Ako Castuera, artist & animator #MarchForScienceLA
8
209
557
For decades, the US government has painstakingly kept American science #1 globally—and every facet of American life has improved because of it. The internet? Flu shot? Ozempic? All grew out of federally-funded research. Now all that's being dismantled. 1/
37
309
636
Meta spokesperson Joe Osborne disputed our core findings, saying our analysis was flawed & that we’d misunderstood the issue. But shortly thereafter, the Cambodian clickbait actors began complaining in their online forums that Facebook had cracked down on their monetizing pages.
3
106
522
And from the scholarship of experts like @jonathan_c_ong and Jason Vincent A. Cabañes who conducted interviews with clickbait actors in the Philippines about their motivations and tactics in the lead up to the 2016 election of Duterte.
1
104
519
I don't think this is healthy for AI development. With the advent of deep learning, people were already concerned about the collapsing of the field around one technique. Now we're seeing it collapse further around a single AI model. This complete lack of diversity cannot be good.
17
93
568
The @WSJ today fired my former colleague for STANDING UP for press freedom. Words cannot describe how twisted & abhorrent this is, at a time when journalists around the world are being intimidated, detained & killed into silence. In awe of Selina's bravery, strength & leadership.
Today I was fired from the Wall Street Journal. This is my statement:
11
165
547
After 3.5 wonderful, wonderful years @techreview, March 30 will be my last day. I still have one more Big Thing coming out in a few weeks, stay tuned. But until then, so much love and gratitude to my colleagues past and present who made TR what it is and me who I am today. ❤️.
46
8
586
Why was Google's firing of @timnitGebru & @mmitchell_ai such a pivotal moment for AI? I spent three months tracing the field's past, present, and future to explain why it matters.
6
168
528
One of the countries hardest hit by all of this platform-funded clickbait activity? Myanmar. Which the outgoing UN special envoy said in October had deteriorated into civil war.
2
125
488
Ever since Google fired @timnitGebru & @mmitchell_ai, it's continued to deploy the very technology it punished them for scrutinizing. Now hundreds of scientists are racing to investigate the technology's risks before it's too late to avoid its harms.
7
205
531