Democracy Now! Audio - Democracy Now! 2026-08-13 Thursday

Episode Date: August 13, 2026

Headlines for August 13, 2026; Trump’s Secret Plane Switch: An Israeli Tip, CIA Doubts & Reporters Put at Risk; “Anatomy of an AI Kill Chain”: Militaries Rely on Mistake-Prone AI in Ukraine, Gaza & Iran; “Deep Unlearning”: Timnit Gebru on AI Hype, Ethics & Algorithmic Racial Bias

From London and New York, this is Democracy Now. In short, we will win this race by becoming an AI-first war-fighting force across all domains. From the back offices of the Pentagon to the tactical edge on the front lines. As the U.S. rapidly expands its use of artificial intelligence, we'll look at how AI is changing the way wars are fought. We'll speak with Heidi Klopp, who co-authored the new report, anatomy of an AI kill chain. We'll also speak to Timnit Geproo, who co-founded the group Black in AI. She was fired from Google in 2020 for warning about biases built into AI models.

But first, we look at the growing controversy over President Trump's secret plane switch in Turkey. did the move put the lives of the White House press corps in danger? Why was the president? On the midterm, you to fly on Air Force One, but not too dangerous for the press. The plane that I flew on was at greater risk. I think it was... Two, we'll speak to the Washington Post reporter who broke this story, all thought and more coming up.

Welcome to Democracy Now, Democracy Now.org, the Warren Peace Report. I'm Amy Goodman. broadcasting today from London. In Yemen, fighters with the Houthi movement have escalated their offensive against government forces with fierce clashes in the province of Thais. The fightings left dozens of Yemeni government troops dead in recent weeks. That's reportedly prompted Saudi Arabia to consider support for a Yemeni government counteroffensive to retake the Houthi-held Red Sea coast.

The renewed fighting in Yemen comes as an unclassified. Pentagon assessment of civilian casualties found the U.S. killed 153 civilians in Yemen last year while wounding another 243. Meanwhile, military families are demanding congressional action over a growing mental health crisis aboard the USS, Abraham Lincoln. After at least six incidents in which sailors attempted to jump overboard amidst an extended deployment, the Lincoln has, has been at sea for over 208 consecutive days, a modern-day naval record while supporting U.S. and Israeli attacks on Iran. Families say sailors have been forced to endure excruciatingly long shifts and shortages of food and basic supplies like toothpaste and soap. Meanwhile, former U.S. officials

say the CIA had low confidence in an Israeli claim that Iran was plotting to assassinate President Trump last month as he returned to the U.S. from the NATO summit in Turkey. Despite that finding, the Secret Service executed a covert deception operation that saw Trump smuggled off of Air Force One and onto a smaller military aircraft aboard a catering truck. The press corps was left on what the Secret Service thought was the targeted plane. After headlines, we'll speak with Dan Lamoff, the Washington Post journalist who broke the story. Firefighters in Lebanon are accusing the Israeli military of starting wildfires in the south of Lebanon. The Israeli military dropped flares in a wooded area Tuesday, and when firefighters came to extinguish the flames, a drone struck nearby forcing them to withdraw.

According to video and testimony from emergency services in Lebanon, the Israeli militaries caused fires in olive groves and on agricultural lands since the ceasefire between Israel and Hezbollah. back in June. In northern Gaza, an Israeli airstrike killed one person and wounded five others. Gaza's civil defense said the strike targeted a Tok-tuk transport vehicle. Meanwhile, MSF, Doctors Without Borders, is claiming Israel is blocking crucial supplies from entering Gaza. These supplies keep hospital generators, ambulances, water pumps, and water trucks running. MSF's head of mission in Palestine said, quote,

what we're witnessing in the Gaza is not a series of isolated shortages. It's a policy of attrition by design. For more than two years, we've observed Israel's policies restricting the entry of supplies. These policies do not merely delay aid. They dismantle the very systems people rely on to survive, unquote. Meanwhile, in the occupied West Bank, dozens of settlers besiege. Three, Palestinian families in Kusra. cutting off water, power, and essential supplies for a fourth day.

A top humanitarian official at the United Nations Security Council Tuesday said conditions in the occupied West Bank have reached a breaking point. This is a Palestinian whose house has been besieged by settlers. The settlers established an outpost close to the house around six meters away. Since that day, we've been subjected to daily attacks by the settlers, night and day, and harassment on the road. They tried to cut off water and electricity networks many times. We fix it and they cut it off again two days later and attack the workers who come do the maintenance. Colombia has authorized joint military operations with the United States against armed Colombian groups following the inauguration of the far-right Trump ally, President Abelardo de la Espreyea.

On Wednesday, U.S. Defense Secretary Pete Hegseth welcomed Colombia to the U.S. Counter-Narcotics Coalition, known as the America's Counter-C Cartel Coalition, or A3C. Hegseth warned the International Criminal Court could investigate U.S. attacks on alleged drug smuggling votes in the Caribbean, which human rights groups have condemned as extrajudicial killings and murder and cold on partners to reject the ICC. That's why I strongly encourage every A-Triple-C member to leave the ICCCC. and reject their attempts to rob your governments and your courts of your sovereignty. I know as a soldier, I swore an oath to the Constitution, not some fake and illegitimate international court. Meanwhile, the New York Times reports fishermen with Ecuador said they're being attacked by English-speaking assailants using drones.

According to the UN, one fishing boat has been lost at sea. since January with eight people still missing and presumed dead. The survivors say the assailants wore matching uniforms with American flag patches. Some of the other fishermen say they've been dragged out of their boats, questioned and handed over to El Salvador Navy before being released. The U.S. Coast Guard and the Pentagon have denied any involvement. To see Democracy Now's coverage of this score, go to DemocracyNow.org. The Trump administration's termination.

federal funding for the Arctic report card, the flagship report, and how the climate crisis is transforming Earth's northern latitudes. NOAA, the National Oceanic and Atmospheric Administration, had issued the report annually for the past 20 years, documenting how human activity is causing the Arctic to warm faster than any other region. In response, former NOAA administrator, Rick Spinnard, blasted the Trump administration for, quote-unquote, imposing ignorance on the system. Dartmouth College professor and polar researcher Melody Brown-Berkins told the Washington Post, quote, it's like not allowing your primary care doctor to how you're doing. Taking away the checkup doesn't stop the changes from happening, she said.

Supreme Court Justice Samuel Alito gained up to $2.9 million. from his fossil fuel interests between 2005 and 2024. That's according to a review of financial disclosures from the nonprofit advocacy group court accountability. This comes as the Supreme Court's about to hear a case with two oil companies, Suncorr Energy and Exxon. These companies want the court to rule that federal law prevents state governments from suing them over climate damage caused by their products.

Activists are calling Justice Alito to recuse himself, citing his extensive investments in energy companies. They're also demanding an investigation by the Senate Judiciary Committee. President Trump announced Wednesday, White House press secretary, Carolyn Levitt, will leave her post at the end of the month. Leavitt was just 25 years old when Trump named her press secretary in November. 24, making her the youngest press secretary in U.S. history. Trump said she's leaving the role to spend more time with her family after giving birth to her second child. The intercept and the Freedom of

the Press Foundation filed a lawsuit Wednesday against President Trump and his staff, seeking to halt plans by Trump's media company to monetize the president's social media posts. Under the scheme, Trump's truth social platform would charge up to a hundred $100,000 per month for early access to the president's posts, which frequently drives stock market volatility. In announcing the lawsuit, Ben Yusig, editor-in-chief of the intercept, said, quote, Trump is trying to enrich himself by privatizing government information that he has no right to sell. We won't let it stand, he said. This comes as the Trump Media and Technology Corporation, which owns Truth Social, announced a $238 million loss during the second quarter of this year.

Meanwhile, the New York Times reports the number of monthly visitors to Truth Social was down about 36% from a year ago. Federal agents arrested the Southern Poverty Law Center's former chief financial officer Heidi Byrick and Wednesday, federal prosecutors charge you with conspiracy to commit wire fraud, conspiracy to submit false statements to a federally insured bank and conspiracy to commit concealment money laundering. Back in April, the Justice Department had charged the SPLC over its now defunct program in which paid informants to infiltrate white supremacist groups in order to monitor their activities. Byrick's lawyer said in a statement, quote, A free and fair society does not use the justice system to silence its political opponents.

Heidi Byrick has dedicated her life to fighting hate groups and extremist movements like the KKK, neo-Nazis, and other white supremacists. Her decades-long record of success dismantling hate groups and the resulting threats to her life speak volumes. For this, she has been indicted, her lawyer said. In North Carolina, the daughter of a pastor who was jailed by ICE, immigration and customs enforcement, says her father suffered stroke-like symptoms that left him partially paralyzed just weeks after his arrest. Gabri Johnson says she's only been briefly able to speak with her 55-year-old father, Gabriel Johnson, since he was taken to a Texas hospital July 30th.

In a now viral video, she said she's worried her father's condition continues to worsen due to medical. neglect. This is not love. This is not care. This is not America. It's torture. And they're trying to not only take people out of their homes and breakup families. They're trying to kill them on the way there. Please, if anybody is within the state of Texas or works in immigration law and reform, I will do anything. Just please help bring my father back home to carry North Carolina. In more immigration news, an ICE agent film pushing an elderly activist to the ground near Chicago last year is pleaded guilty to a single misdemeanor battery charge. Adam Seraco was sentenced Wednesday to 12 months of court supervision for the assault.

In December, he'd recently completed a shift and was off duty when he assaulted 68-year-old attorney Robert Held in a gas station parking lot. Held had been recording video of Seraco filling his gas tank. Meanwhile, a widely circulated video shows an ice agent in full. Falls Church, Virginia, pointing a gun at a driver who yelled at officers as they detained people on a sidewalk nearby. Video filmed by the driver shows masked ICE agents repeatedly accusing her of trying to run them over with her car and threatened to arrest her. When she contradicted their claims and said dash camera video would vindicate her, the agents walked away. And those are some of the

headlines. This is Democracy Now, Democracy Now.org, the War and Peace Report. I'm Amy Goodman in London with Nourmet, Sheikh in New York. Hi, Nirmie. Hi, Amy. And welcome to our listeners and viewers around the country and the world. We begin today's show looking at how an alleged Iranian threat to assassinate President Trump led to a secret operation last month that possibly endangered the lives of journalists and White House staffers. The Washington Post reported on Monday that President Trump secretly switched planes before departing a NATO summit in Turkey. Ahead of his departure from Ankara on July 8th, President Trump announced on truth social that he would not be flying on the newly refurbished Boeing 747-8 jet gifted by Qatar, but instead

on the older Air Force One, for old times' sake, he said. That announcement led to revelations that the Qatari plane lacks some of the safety features of the Legacy Air Force One, leaks that angered the White House and led to subpoenas of New York Times reporters. But the Post revealed Monday, the president actually departed on a third plane. President Trump was seen boarding the old Air Force One, but then secretly left through another door and was shuttled in an airport catering truck, along with several a third plane. The press was asked to lower their shades before the U.S. the operation presumably so they would not see the president's exit.

Senior officials who were left on the plane along with the staff and reporters included Secretary of State Marco Rubio, Treasury Secretary Scott Besson. The White House never revealed the secret operation. On Tuesday, President Trump was asked by reporters about the plane switch and the risk posed to the reporters on the decoy flight. Secret Service. I just follow what they'd like to do. So I go by secret service and the military. Why was the President on the midterm for you to fly on Air Force One, but not too dangerous for the press? The plane that I flew on was at greater risk. I think it was a clear. Mr. President,

that would be the plane, I think, that they would be more likely to go for it. White House Correspondence Association President Jackie Heinrich of Fox News met privately with senior White House officials Tuesday to convey concerns from the press corps. In an internal memo to members afterwards, she said, quote, I asked the White House to work with us on a protocol that recognizes the Secret Service's responsibility to respond to genuine threats while protecting press safety, preserving independent coverage of the president, and providing a mechanism to correct the record after the fact when necessary.

For more, we're joined by the reporter who broke the story, National Security writer for the Washington Post, Dan Lamoth. But his latest piece is headline CIA had low confidence in Iranian threat before Trump switched planes in Turkey. In it, Lamoth reports the intelligence about the assassination threat against President Trump had come to the CIA from the Israeli government, but was not seen by the CIA as high. highly credible. Dan, before you talk about the president believing he was threatened and moving to a third plane, first the cutter jet, then the old Air Force one, and then this military jet, before you talk about him leaving it and presumably leaving the press in the endangered plane,

you now have this new story that the U.S. CIA didn't even think Israeli intelligence, because that was the source of the information, about the assassination threat to Trump was viable? That's right. We reported last night, my colleagues, Warren Strobel, John Hudson and I, combine on a story that kind of laid out how they came to the decision to make these extraordinary moves in July.

The Israelis passed a tip, a concern that the plane and potentially the president was under threat. The CIA had low confidence in it. And I think that kind of leaves the secret service in a tough spot of assessing what do you do in this zero fail mission to keep the president safe, but at the same time, you know, manage the additional pieces. They went with the more extreme option of removing the president and then keeping its secret after. So Dan Lamoth, if you could put this in also

historical context. In other words, in the past, the Secret Service has used decoy, motorcades, and planes for presidents. Why is this incident different? On previous occasions, were the press corps and other senior officials also taken off the plane? The most direct example we've seen is probably President Clinton landing in Pakistan in 2000. The Post had another piece earlier this week from Scott Nover, where it was acknowledged and reported that at least the chief of the White House Correspondents Association at the time had been brought in quietly. They kind of laid out what they needed to do or what they thought they needed to do. And in that case, the president followed in Air Force One on a second plane. But I think the most significant difference here is that

And in that case, moments after landing, President Clinton popped out of the other plane, and it became very apparent what had happened. So, you know, within minutes, it becomes clear. Everybody knows there's a decoy. It was reported at the time. Basically, what happened and why. You know, and in this case, you've got, you know, it was handled very differently. I mean, a month worth of secrecy, presumably that would have continued had it not been for some news report.

Hearing a few of thoughts on, I mean, here the White House subpoenaed New York Times reporters. They were the ones who exposed the lack of whatever, enough defense on the cuttery plane that was gifted to Trump. So all the attention in these last weeks has been on that. They were ultimately forced to drop those subpoenas. But maybe the White House was most concerned that it was this that would come out, that he left the jet, left the reporters on. on the jet and moved on to a third plane. And yes, Marco Rubio was apparently on that plane, as was Scott Bessent. He was with his personal aides like Walt Nauta, who was indicted along with him

a few years ago, though those indictments were dropped. Your thoughts? Yeah, I mean, when this tip came to me, I was struck by the extraordinary nature of it. Initially, kind of looked at it as like this seems kind of surreal. And if we're going to report this out, we better be damn sure before we publish. So it took time to run this down. Multiple conversations, multiple source interviews. And I thought really helpful here. You know, that initial story, we got our visual forensics team involved. We got our investigations team involved. We comb through airport tracker data. And really, I thought one of the remarks things in that initial story.

And something they made it hard to deny was you had video in the airport that basically followed a lot of this right through. If you're looking at that catering truck as it's just another random day, it didn't stick out. But when you look through it through the lens of the other reporting, it looked pretty extraordinary. And Dan, finally, on another topic, you've been reporting on impacts of the Iran conflict on munitions, military personnel, and on, you.

U.S. facilities in the region. Could you talk about some of that reporting and what you found? Yeah, I mean, really, this has been an enduring concern throughout this conflict that really grows by the day as the conflict kind of meanders on. United States has a large number of munitions broadly. The problem here is they do not have enough of very specific and very important munitions. So over time, things like Tomahawk missiles, Patriot interceptor missiles, things that have really played a key role in this particular conflict and have also been important in the Ukraine conflict and would be important in any kind of conflict with China or some other adversary are dwindling. And they cannot be built quickly. And this was a known thing going back not only through the Trump administration, but into the Biden administration as well. some steps were taken to fix this.

However, once they haven't been able to really resolve that issue, and really they've just tried to downplay it. Dan LeMoth, we want to thank you for being with us. National Security Writer at the Washington Post will link to your exclusive that exposed the secret flight is headline contradicting public statements. Trump took secret flight from Turkey amidst Iranian threat. his latest piece with other Washington Post reporter, CIA had low confidence in Iranian threat before Trump switched planes in Turkey. Coming up as the U.S. rapidly expands its use of artificial intelligence. We'll look at how AI is changing the way wars are fought. We'll speak to Heidi Kloff.

She co-authored the new report, anatomy of an AI kill chain. And then we'll talk about race and AI. Stay with us. Swagut. Yes, I am here in London, and that's exactly what Londoners and people in so many countries in Europe experienced yesterday a solar eclipse. This is Democracy Now. Democracy Now.org, the Warren Peace Report. I'm Amy Goodman in London with Nermin-Sheikh in New York. We turn now to how artificial intelligence is transforming warfare from targeting software to autonomous drones. Earlier this year, Defense Secretary Pete Hexeth spoke at Elon Musk's company SpaceX. In short, we will win this race by becoming an AI-first warfighting force across all domains. From the back offices of the Pentagon to the tactical edge on the front lines.

And this is Sentcom Commander, after. Admiral Brad Cooper talking about the U.S. military's use of AI. Our war fighters are leveraging a variety of advanced AI tools. These systems help us sift through vast amounts of data in seconds so our leaders can cut through the noise and make smarter decisions faster than the enemy can react. Humans will always make final decisions on what to shoot and what not to shoot and when to shoot, but advanced AI tools can turn processes that used to take hours and sometimes even days into seconds.

We're joined now by Heidi Klaff, the chief AI scientist at the AI Now Institute. It's just released a joint report with Air Wars titled Anatomy of an AI kill chain. Heidi, thanks so much for being with us. We want you to explain this report, but also to warn you to a global lay audience when you use terms like generative AI and artificial general intelligence, if you would explain your terms. But tell us about this anatomy of an AI kill chain. Yes, we actually wanted to break down and demystify a lot of these terms that we're now seeing being used, like you mentioned, generative AI, large language models, and also more

generally the use of AI generally in warfare and what that means. And I think it's important to remember and what we're trying to convey in this report. really, is that AI has been used in the military defense since the 1960s. And when we're talking about a new AI arms race, it means the inclusion of something like chat GPT, which is a type of AI that's known as a large language model into now the decision-making processes for targeting. Right. And the thing is that we want to break down is that it's not just one AI killer robot, which is often kind of the concept that people have of AI. What it is is a series of different types of AI algorithms that are chained together, each with their own pitfalls, each with

their own flaws, each have very low reliability and accuracy. And these systems build off of each other and feed into one another to unfortunately end up in resulting in civilian casualties, as we're often seen with their use. So we break down each step of the kill chain, how each AI algorithm leads to faulty decisions and ultimately can lead to decision-making in targeting civilian casualties because of how faulty these algorithms are. We don't just focus on the new types of AI that we are seen advertised, you know, as discussed things like chat GPT, large language model. We also talk about older types of AI that are continued to be used and continue to be relied on, like vision systems. And if you could explain, Heidi, the difference between

decision support systems and autonomous weapons systems. And on the question of large language models, you've warned that there have been operations from Russia and China that put out propaganda to try to skew the outputs of large language models. What does that mean? So really the difference between a decision support system is an autonomous weapon systems is that there's a human in the loop when you're selecting a target, essentially, to strike. An autonomous weapon system means that this is done automatically without a human in the loop. And I considered this kind of, you know, separation between them to be pretty superficial in practice. Because ultimately, if you have a human who is just rubber-stamping the AI decisions, right, and we know that they do this due to kind of something called automation bias in our field, which is that we know that humans are prone to just accepting recommendations of AI algorithms without corroborating that evidence. To me, in practice, that means

that you have a human who under the time pressures of war and conflict is just approving decisions by an AI. And so typically this separation is often made to say, more the AI is safe, it's fine even if it makes mistakes because you have a human correcting that. But in practice, we see a very, very different photo or a different picture of how that occurs. You could explain the way in which AI systems were used both by Israel and Gaza as well as in the U.S. Israeli war on Iran,

and in particular the attack. on the school in Minab in the first days of the war? Yeah, I mean, we've known that large language models have been used since really the idea I started using them in Gaza. A good example of that is an AP investigation, which demonstrated that they were using it to translate intercepted communications, right? And they were taking essentially Arabic and determining whether or not specific people should be put on a target list.

And we found essentially that because, again, large language models are highly inaccurate and unreliable, that they mistranslate Arabic. In this case, they took something like a payment and thought it referred to a payload and added people to a targeting list. In this specific example, we know that this was caught. But again, under the time pressure, there's going to be oftentimes situations where human operators aren't going to double check that, especially that oftentimes it's very difficult to understand why a specific AI made the decision that it did.

And now we know that it's been confirmed by the U.S. Department of War that they are using Claude through Palantier's Maven to similarly make targeting decisions. And we have now, unfortunately, the Minab School tragedy, which resulted in 160, you know, civilian casualties. And it's becoming then unclear whether or not AI is used that we know that Cloud is being used to sort of make these targeting recommendations. But ultimately, it hasn't really been clear. Was it the AI, was it not? Was it deliberate? Was it not? And how do we really trace that when we can't really verify or validate whether something was an AI recommendation or not? Furthermore, we have a very difficult time with large language models because of their scale and nature of operation to be

able to understand why an AI made the decision that it did. Was it due to faulty data? Was it directed to do so? Was it due to hallucinations? Which, if anyone has used chat GPT, knows that these you know, chatbots often kind of fabricate information. And I think that's where we are with that current situation. It's so unclear to us what was the cause of that. And I would say actually, that's kind of the purpose of AI system themselves. They are often used to launder accountability. Heidi Kloff, Anthropic earlier this year announced it'd be ditching its core safety promise.

You previously worked at companies like OpenAI, where you tried to help them create their first safety frameworks to understand how to evaluate their AI systems. And now OpenAI's head of ethics left after less than a year after joining. That was Chloe Baccarlyar, one of several high-profile exits at Open AI in recent weeks as concerns mounted. If you can talk about what these ethics frameworks, are and why people are leaving one after and other and why it is, it's the only the company that determines this, not government regulators, how AI is used. Well, I can't necessarily speak on why those specific individuals left, but I can speak

from my own experience as a safety engineer because I, prior to join an open AI, I actually worked in nuclear defense and aviation, ensuring that the systems implemented in those, you know, what we call safety critical situations don't cause any harm. And ultimately, what these companies are, are they're reinventing safety. They're co-opting a lot of the terms that we've used in traditionally in safety and security engineering to really make this about kind of a different types of safety. Instead of being concerned, do our systems, are our systems reliable? Are they accurate?

Do they perform as intended? Are they fit for purpose so that no human is harmed and sort of due to the decisions that they or recommendations that they make. We instead have these like safety and ethics frameworks about, you know, whether or not these systems are autonomous, whether they have intention, whether they're conscious, right? And that's a very, very different question. And often they can give the illusion that they are and sort of, you know, personifying a lot of the language, you know, or the actions of these models themselves. I mean, we've seen recently, you know, these, the cybersecurity capabilities being conveyed as kind of like, oh, this is autonomous. This is something we have been expected from

these things. They're doing this on their own will. When these companies actually deployed them in a very insecure environment, they intentionally were having these models, you know, go on to sort of carry out cyber attacks as part of sort of testing or benchmarking. And so this is the reason that I take great issue with that is that they actually, again, try to assign accountability to the AI, rather than themselves. Really, I view that AI only does what we tell it to do is only capable of what we build it to do. And these AI companies are, again, moving away from accountability by trying to assign sort of autonomy or personify these AI themselves in doing that. And I think, you know, when you're then deploying these kind of models in warfare, it's very, very easy to then sort of say,

well, it wasn't us who made the decision. It was AI. And that's why I take great issue with the way they're now talking about safety and security engineering. And indeed, as you say, I mean, now it's very routine to hear about AI agents, so to speak, going rogue. And yesterday there was a report that Taiwan was hit by an especially abnormal AI-assisted cyber attack, suspected to be from China. And reportedly, this kind of end-to-end autonomous attack on a government target has never been seen before. And the most remarkable fact about the target, as the Financial Times wrote, is that the AI continuously ranked and reprioritized possible attack plans based on available evidence.

So if you could talk about whether you think in all of these cases, or the ones that we know about at least, it is the makers of this AI who are responsible and not the agents that are going rogue. and what it would mean if these agents, artificial intelligence agents, move beyond cyber attacks and into actually theaters of war? Well, the thing is we are seeing them being deployed in warfare already. And the thing about AI is that it is a very powerful technology. They are very capable things, but they're only as capable as we make them. And they only really do the things that we make them do. And oftentimes we might not be precise in the way that we specify.

their tasks, right? Like the cybersecurity tasks, they were assigned to go and carry out cybersecurity attacks. You know, they were not doing this on their own will. They were built that way. They were trained that way. And I think that's very important to keep in mind. And in warfare, it's the same situation. They are being basically given a lot of data to understand how to target. They are being giving profiles often faulty ones to determine who's an adversarial and who isn't. And ultimately, the thing is, these are based on really loose parameters, right? These are not, the way that we build them isn't always accurate, isn't always reliable, and we know that AI really doesn't really understand anything beyond what it's trained on.

So as soon as it sees a situation, especially in warfare, that it does not recognize, we know this is in the exact situation that the AI will fail, make a recommendation, that's not true, or fabricate outputs, right? And so I think it's really important to keep in mind, yes, they are capable. Yes, they can do things like cybersecurity attacks. and they do speed up operations. They definitely do. That's kind of, you know, one of their core features.

But then if you have something that speeds up operations, but their kind of baseline accuracy and reliability is 30 to 60 percent, that's not really different from indiscriminate targeting, right? It doesn't also mean this is some sort of, you know, killer robot that has its own intention. Anything has been built by design. And if we're reckless in the way that we build them, that doesn't mean it's the AI is at fault.

I think it definitely should that counterfeit. should always be with the system designers and the people choosing to deploy them, despite knowing that they have these really poor reliability and accuracy rates, especially in new situations, which is actually more likely to happen in warfare than it is with sort of cybersecurity attacks. Heidi Kloff, we want to thank you for being with us, Chief AI scientists at the AI Now Institute. We will link to your report that you released jointly with Air Wars titled Anatomy of an AI kill chain. Coming up, Timit Gibrew, the founder of the distributed artificial intelligence

research fired from Google in 2020 for warning about biases built into AI models. Stay with us. Matter by the corner laughers. This is Democracy Now, Democracy Now.org. I'm Amy Goodman in London with Nirmine Sheikh in New York. We're continuing to look at how artificial intelligence is rapidly changing the world. We're joined now by a leader in the field of AI ethics, Timnit Gabru. She is the founder and executive director of the Distributed Artificial Intelligence Research Institute, or Dare. Prior to that, she served as co-lead of the Ethical AI Research Team at Google. She was fired in 2020 for writing a paper warning about the dangers of large language models and raising issues of discrimination in the workplace.

Tim Needs also co-founded Black in AI, a nonprofit that works to increase the presence, inclusion, visibility, and health of black people in the field of AI. Her forthcoming book is titled Deep Unlearning, The Rise of AI and the Radicalization of a Tech Idealist. She joins us now from Boston, Massachusetts. Tim Neat, could you begin by explaining what AI means to you? You've said you think of AI as a discipline with a number. of sub-specialties. Explain. Yes, and I'm glad that Heidi did some groundwork for us here in the earlier segment.

But, yes, for me, AI is a discipline with a collection of specialties that are all lumped into this term AI, and they can be very different. So Heidi mentioned things like large language models, but also computer vision systems that have been around since the Vietnamese. war. And all of these things are bundled into the term AI. And in fact, the things that are considered AI are not always considered AI. So, for example, the title of my forthcoming book is Deep Unlearning. And this is a play on the term deep learning, which is a technique, which is currently basically synonymous with AI. But back in the day, deep learning was not considered AI so much

so that the researchers in that specialty were having their academic papers rejected by the prestigious AI conferences. And so actually deep learning was sort of a rebranding from another term artificial neural networks to avoid that stigma. So not only do we have this marketing term that is a hodgepodge of disparate techniques or tools.

And in fact, it's not even sometimes it's techniques, sometimes it's tools, sometimes it's products. But also, not all of these things are always called AI, right? When one of these subspecialty sort of is really hyped up or has some sort of better performance on a specific benchmark, it gets elevated to AI. And then it gets sort of downgraded not to AI. And there are other such examples like expert systems of the 80s. That was what was synonymous with AI in the 80s. And now nobody calls expert systems AI.

So what happens here is that it makes it so difficult to have conversations that are grounded and specific and have specific conversations about what are the harms, what are the benefits, what approaches are good, what approaches are bad? Because when I talk about some things that I'm really against, like chat GPT kind of things, then people are like, what about a particular system that is used for early diagnosis of cancer patients, you know? And I have to say, well, that's a completely different system. It doesn't even have to be built in the same way that these big models are being built. Tim Nizkebrou, can you talk about co-founding Black and AI? But start off by talking about why you were fired from Google in 2020, the issues that you were raising there and the position you had there.

So yeah, I co-founded Black and AI way before I even joined Google. My current institute is called the Distributed AI Research Institute, which I found it after I was fired from Google. But I started Black and AI in, you know, around 2016, 2017. When I attended, you know, I started seeing simultaneously the lack of black people in AI. And so I would go to academic conferences and you'd have about five, six thousand people in those conferences, and only a couple of one or two, a handful of black people. And at the same time, we had some of these systems, some of the kinds of systems that Heidi was talking about.

For instance, there was a purportable car article from 2016 talking about a company purporting to determine someone's likelihood of committing a crime again. And so this company was saying that they built a software. that can tell whether a person released from prison was likely to commit a crime again. And the ProPublica article was talking about how these systems were more likely even to label black people as criminals, potential criminals, than white people. And it was very scary because judges were already using the outputs of these models in their decisions about bail, decisions about how long someone should be imprisoned for. So the

dichotomy of seeing the lack of black people in the field and also the kinds of claims that were being made and the kinds of things these tools were being built for was very scary. And so that's kind of how I decided to found black and AI. So by the time I started working at Google in 2018, I was a very known quantity. I had already, you know, founded and led black. I was working on uncovering all sorts of issues in the field. My collaborator, Joy Buwamini and I had written a paper that was pretty, made the rounds, was highlighted all over the place and even changed policy, showing for the first time that automated facial analysis tools like face recognition, for instance, tools,

were, had much higher error rates for darker skin. women than lighter skin men. And the darker and darker the skin, and especially for women, the higher and higher the error rate. So think about these tools being used on CCTV cameras to identify so-called criminals, et cetera, and the people who are of darker skin would be more likely to be wrongfully identified in this case. And so I had already worked on this kind of these kinds of uncovering these kinds of issues. And so I was hired at Google in 2020. to co-lead a team called the Ethical AI Research Team, which was a small research team with Meg Mitchell, who was also later fired.

And there were many issues. I mean, it was one issue after another. This was in the middle of the Google walkout where 20,000 women walked out because we found out that Google had paid Andy Rubin $91 million after allegations of sexual misconduct. And so it was a combination of a whole bunch of issues, and it was at the height of Black Lives Matter, the Black Lives Matter movement in 2020. And at this time, a company called OpenAI that was getting very famous, but not as famous as it is today, came out with a large language model called GPT3, which later became the backbone of chat GPT. but chat GPT wasn't released yet. And GPD3 was really hyped up in the mainstream media.

And OpenAa was making all sorts of claims about how powerful GPT3 was. In fact, actually, they had said that its predecessor GPT2 was too dangerous to release because it's so powerful. And so the GPT3 was a large language model. we can just define what large language models are. They are models that are trained on vast amounts of textual data on the Internet, and they're trained to calculate the most likely sequences of text based on the training data.

So if you use large language models to generate text, you are kind of trying to generate the most likely sequences of text given your training data. So the fact that opening I got so much airtime for GPT3, meant that all of these companies wanted to build similar models, and they wanted to build larger and larger models, which means that they were guzzling all of the data on the Internet, and they were consuming, using huge amounts of computational power even then. And so we were very worried about this race that was started.

Every company wanted to have the largest model, and we were asking, why do we need the largest of any? what problem are you trying to solve? And we warned about, we wrote a paper, my collaborators and I, warning about the dangers of large language models, and we said, how big can language models be too big? So the first issue that we mentioned, the very first one, was the environmental and financial cost.

And you can see how six years later now, that is not only true, but even worse than we said, we said in the paper. We talked about how the carbon footprint of these huge data, the necessary computational power necessitates huge carbon footprint. And even if you claim to have data centers with renewal of bill energy, that energy is going away from heating people's homes, for example,

and towards training these models. And the financial cost that shuts out anybody who doesn't have the wealth to participate in this kind of work. we also warned about perpetuating hegemonic views. The claim was that because these models are large, which means that they are using huge datasets, then, oh, they have all of human knowledge in these data sets. But that's not true.

The Internet represents hegemonic views. It does not represent views of everybody in the world. A lot of people are not even on the Internet. We even know articles like Wikipedia are. heavily biased. It's overwhelmingly Western and male. And we also warned about, and this is very related to what Heidi was talking about earlier, we talked about how because large language models are, tend to output fluent and coherent text, this can be very deceiving. So one example we gave was from 2017. And so these large language models, you know, they could be used. They were

used as systems in a whole bunch of other things like machine translation systems. And in 2017, a Facebook translate translated Palestinians, Good Morning into attack them. And, you know, attack them. There's no grammatical error or anything like that. So it doesn't even give you a cue that this translation might be incorrect. And this person was arrested and later released. And they didn't even check to see what the untranslated version was when they arrested him. And we call this automation bias, the tendency to over-trust the outputs of automated systems. And so we warned about also people interacting with these kinds of texts output by these kinds of systems, attributing a mind behind whatever text they're seeing. And so, you know, right now, the, the, these chatbots

like Claude or ChatGPT are designed to make you believe that there's some sort of superhuman brain behind whatever is outputting these texts. And that's, that's hugely dangerous. And even though we weren't speaking about chatbots back then, we were already talking about the tendency of people to attribute a mind behind these texts, and so which means that, you know, they can be deceived into believing that there's something, that something more than a model outputting text. And the final thing we warned about was,

what is the cost of pursuing this singular direction in terms of research and resources? What is the cost of shutting out all other possible futures, all other ways of building machine learning systems, which is a sub-specialty of artificial intelligence or any other kind of paradigm, because it seemed like the whole field was only investing in this singular direction of building larger and larger models that are guzzling all of the data. I mean, stealing so many people's works and based on exploited labor and killing the environment.

And so, of course, you know, I got fired after that. I mean, there's details, but that was the icing on the cake. It was writing that paper. So, to me, you know, we only have a couple of minutes left, but just to point out that, you know, the scale of which these AI bots are being used, very recently, chat GPT crossed a billion active users a month. Google Gemini has said now that it has done the same. But I want you very quickly to talk about the work that you do at DARE,

in particular with people. working in these data centers and the conditions under which they work around the world, mostly in the global South? Yeah, we have a project called a Data Workers Inquiry Project, and this one is working with data workers, and these are the people that label the data that is used in these models, which is why we think that they're super intelligent, right? These people have to painstakingly supply the data and label it, and they are, work in overwhelmingly exploitative conditions. They're paid maybe in $1 an hour.

They don't have any breaks. I mean, and so we have this project that you can look at where they perform research into their conditions. They organize. We help them create an organization called the Data Lablers Association in Kenya. And so for me, it's really important to make sure that the voices of the people who are negatively

impacted our elevated and they're the ones telling their stories. And it also shows us that we don't have to do things. This doesn't have to be the way. The path we're on right now was never a preordained path that we had to be on. There were many other possible paths that we could have taken. Timnit, I wanted to ask, you grew up in Addisababa in Ethiopia. Your dad was an electric engineer, your mom, an economist. What led you, coming to, as a to the United States to get so deeply involved with the leading mind on AI right now. We just have a minute. Yeah, I mean, actually, my book kind of talks about that journey because I want people

to understand that tech does not have to be the way it is right now. But I always loved math and physics. And I also thought it was a refuge from the messy world of war, which is what made me leave Ethiopian politics. But I later learned that that's not true, actually. The war in politics that I left is what drives the world of math and science and technology as well. And if we don't understand that, we'll continue to harm our own people and as engineers and scientists build terrible and harmful products instead of ones that actually support our communities. Tim Nietjeburu, want to thank you so much for being with us, founder and executive director of the

Distributed Artificial Intelligence Research, that's Dare Institute. We look forward to interviewing you on your forthcoming book titled Deep Unlearning, The Rise of AI and the Radicalization of a Tech Idealist. She also co-founded Black and AI. She previously served as co-lead of the Ethical AI Research Team at Google, fired in 2020, for writing a paper warning about the dangers of large language, models and raising issues of discrimination in the workplace. That does it for our show.

Special thanks to our crew here in London, Julian Jones, Pablo Delbracio, Grace Garrett, Dennis Moynihan, Hannah Elias, and the whole team in New York. I'll be next week in Middlebury, Vermont, and then we're on to Madison, Wisconsin, and Chicago. I'm Amy Goodman in London with Nirmine Sheikh in New York.