What if AI could be a tool for freedom? In conversation with Owen Clegg from BBC World Service, Alex Gladstein, Chief Strategy Officer at the Human Rights Foundation, explores how artificial intelligence can empower human rights activists, expand access to information, and help counter authoritarian uses of technology.
The World Service reaches 355 million weekly listeners across more than 100 countries, with a large audience listening from inside authoritarian countries.
While AI is often framed through the lens of risk, Alex explores how the technology can also empower activists, journalists, and opposition movements around the world by supercharging their work through personal and organizational agents. He points to Venezuela, where opposition groups responding to a humanitarian crisis have used AI to quickly build crowdfunding infrastructure that would otherwise have required significant time, technical expertise, and resources helping deliver aid. Listen to the full conversation below.
Transcript
Read the full transcript
OC: We are going to talk now, though, about artificial intelligence. The headlines last week were dominated by the story of that AI system that went rogue and hacked into another company. We’re going to hear now, though, from somebody who says that AI can be used for good. Alex Gladstein, the chief strategy officer at the Human Rights Foundation, says that the power of artificial intelligence can be harnessed as a tool for liberty. I spoke to him earlier from California.
AG: I think people might be surprised to hear that AI is awesome for human rights. And in order to get exactly why, let me just break it down. It gives human rights activists superpowers. So if you consider dictators, they already have infinite numbers of law firms, top research units, design studios at their fingertips. They have all these resources. But who’s actually making human rights happen? Who’s advancing human rights on the front lines? It’s journalists, opposition groups in difficult political environments, labor and environmental groups. And these organizations or individuals are typically underfunded. They don’t have a lot of resources. So AI agents put essentially law firms and design firms and top research into their pockets. And it can dramatically scale the effectiveness of their work. I mean potentially, exponentially I mean obviously dictators and governments are using AI but you know they already have a lot of resources. I’ve seen firsthand how this can completely be life-changing for the average human rights activist.
OC: Okay give me an example then of a situation where artificial intelligence has recently changed an individual’s nation’s human rights situation for the better.
AG: Sure, there was just a horrific earthquake in Venezuela as I’m sure you all know. And, you know, a lot of the people trying to work to help the individuals who’ve been hurt are being hindered by the dictatorship there. So some of the opposition groups who normally, when you’re trying to launch a crowdfunding campaign, it would involve hiring a web firm, you know, figuring out a banking portal and all this stuff. A friend of mine who is a very well-known opposition leader there, Lilian Tintori, you know, she was able to use her agent to essentially create this crowdfunding platform in, you know, a matter of hours instead of weeks and weeks and weeks and thousands and thousands of dollars of expenditures. You know, these activists can do these things and they can react to real life events and they can do things by themselves. And it’s extremely liberating.
OC: Do you regard artificial intelligence in the round then as something that is good for society that we will all benefit from.
AG: It completely depends on what kind. So, and this is where you all have seen in the news this week, a huge letter being signed by all the major tech companies in the United States, with the possible exception of Anthropic. And this relates to open weights versus closed AI systems. Open weights are super important. Open weights are just AI models that anybody can download and customize and own. And closed weights are the ones that, for example, Anthropic uses where you have to pay them and you can’t access it yourself. You can’t have it on your own terms. Open weights are the ones that we’re going to need for the human rights struggle. They’re the ones that people can actually control their own intelligence. This is super, super important. So the way you’re going to see this play out is activists are either going to be able to access open-weight models privately through encryption schemes on the internet. You can already do this now through things like Maple and Tinfoil. Or they can use AI locally by downloading the intelligence onto their phones. A lot of people don’t know this, but you already have an intelligence if you have an iPhone. It’s built into your phone by Apple, and it just lives entirely on your phone. Nothing leaves your phone. It’s fully private. And with an app like Atomic Chat, you can download it and start using it as a chatbot without having to go pay ChatGPT and leak all your information to this company.
OC: But these systems still need access to the internet, don’t they? And what we saw during, for example, Iran earlier this year when the mass anti-government protests took place is the government simply flicked a switch and turned off the internet.
AG: So that’s the cool part. The local AI does not need the internet. So if you use Atomic Chat… You’ll have a model that lives on your phone. You can be on airplane mode, and it can help you with complex research and anything that more or less chat GPT can do. It can do sort of 90 cents on the dollar. It’s pretty good for something that’s completely free, lives on your phone, and is fully private. So I think that when we’re talking about the struggle for human rights, we’ve got to think more about local AI and sovereign AI and AI that we control. There’s another app you can download called JAN for your MacBook or your computer. Do the same thing. You can run models at home. And this is just like the personal computing revolution. In the 1970s, people would have to go to the library or their company to use a computer. And then it came into their home and they got to own it. The same thing is happening with intelligence, with these AI models. And it’s only going faster and faster as basically the amount of memory you need to run intelligence is shrinking and shrinking and shrinking.
OC: Have you seen artificial intelligence being used in a way by governments and perhaps in particular authoritarian governments, which has unnerved you?
AG: Well, at the Human Rights Foundation, we’ve supported research into, for example, China, which has in place all the ingredients for a very dystopic AI system. They have hundreds of millions of cameras in public places, they have mobile phone tracking. All of this big data analysis can be used to do pre-crime tracking, sophisticated real-time tracking of ethnic minorities, censorship. But the thing is, you can’t fight a computerized dictator with a typewriter, and you can’t fight an AI-powered dictator without AI. You just can’t. So it’s really important that human rights activists figure out a way to embrace this tool on their terms with their philosophy, where we’re talking about anti-authoritarianism, and intelligence that they control.
OC: You mentioned China there. Of course, anybody using an AI system in China would be doing so within the Chinese internet firewall, wouldn’t they? Or are you saying that these systems could allow people in China, for example, to read about the Tiananmen Square massacre, which they can’t do on the open internet?
AG: Right. So there are ways to get these local models on your phone, and you don’t need the internet to use them. And you could drop these into North Korea if you wanted to. We’ve done a bunch of work there with getting information into that country. It’s like an offline, sophisticated encyclopedia. Essentially, it’s a chatbot that doesn’t need the wider, you know, the World Wide Web. Those could work in any situation with no internet. And, you know, at the end of the day, China’s playing a risky game here with supporting all of these open AI systems, ironically. It’s great for human rights, but in the long term, I don’t think it’s going to be great for the Chinese Communist Party because they’re pushing out into the world systems that people can control. And I understand they’re doing that right now to try and break up America’s economic monopoly over AI. But I think it’s going to come back to bite them. And I think it’s just inarguably good for humans, for open weights to spread around the world, regardless of who makes them.
OC: Alex Gladstein there from the Human Rights Foundation, what he thinks could be a benefit from artificial intelligence.
