What Happens When You Give Millions of People Free Access to AI?

Hundreds of millions of people worldwide use generative AI chatbots. Tens of millions of us pay for access to premium versions of those products — at least 20 million pay for ChatGPT, according to an OpenAI spokesperson.

But a clutch of countries, including one in Europe, are flipping the equation, offering access to AI tools at a population level in the belief that it can kickstart their economy and improve tech literacy.

The furthest along is Sweden, where a homegrown champion is offering AI access to around a quarter of the population through a foundation established at a summit in Stockholm earlier this year by tech investors Olof Hernell and Nicklas Berild Lundblad, and tech founder Joel Hellermark.

Sweden’s AI Reform initiative is co-funded by Sana Labs, Hellermark’s AI unicorn, recently purchased by HR software company Workday for a reported $1.1 billion. The company has matched 12 million Swedish krone ($1.3 million) of private donations to fund access to Sana’s AI agent and chatbot for public sector workers, teachers, students, non-profits and research institutions in Sweden.

The goal is to get 2.3 million Swedes hands-on with AI, bolstering the economy as the 1990s PC Reform initiative did. Then, the Swedish government offered tax breaks for households to buy computers. “It got around 1 million Swedes a PC early, and that’s out of 4.4 million households at the time,” Oscar Werner, CEO of Gilion, an AI-powered funding platform based in Stockholm, told Tech Policy Press. Werner is a native Swede who grew up steeped in the country’s tech sector, and recently returned after a stint in Silicon Valley.

The PC Reform, followed by the deregulation of Sweden’s mobile market, propelled the country to Europe-leading tech adoption and gave birth to Skype, Spotify and Klarna. “If you count the number of unicorns by capita, this is one of the hottest areas in the region,” said Werner.

The aim is to repeat the trick with a potentially even more transformational technology: generative AI.

A year before Sana launched the AI Reform, a Swedish government committee tasked with looking at AI, recommended providing universal access to AI tools to improve equality. “If more companies join in, and if we get the spread and the funding that we envision, I foresee that we would have in five years none of these groups that have fallen behind in the use of AI,” Olivia Elf, head of operations at Sana Labs, told Tech Policy Press. “I think access to AI will be as important as education,” she said. Added Werner: “What Sana’s doing now on the AI side is the same thing [as the PC Reform].”

So far, only “a few thousand” people have taken up the offer, according to Elf. But that’s in part because it’s designed to target those least likely to use AI today. “Our vision with this initiative is to accelerate the groups in society to equip them with AI or access to AI, and especially these groups that might not otherwise start using it as easily,” said Elf.

Among the early adopters is Daniel Vare, a project manager at KTH, Stockholm’s technical university. When we spoke in early October, Vare hadn’t used Sana for a few weeks. He’d maxed out the previous month’s 200-credit allowance on Sora within a week. “It doesn’t matter that I burned through the license, because this is an extra playground,” he told Tech Policy Press. Like many people his age, Vare, who’s in his mid-40s, was a beneficiary of Sweden’s 1990s PC Reform initiative.

Another early adopter is Niklas Ek, business developer and project manager at Generation Pep, a small non-profit founded by the crown prince and princess to encourage healthier lifestyles for Swedish children. “A couple of years ago, we didn't use AI whatsoever, and now most of us are using it every day and for more and more complicated tasks,” said Ek, who was a 16-year-old when his family got their first computer through the PC Reform. “The AI is getting better month by month. It's incredible.”

Vare uses Sana’s chatbot to strike the right tone in emails and schedule reminders for funding deadlines. It’s paid dividends in improving his efficiency. But he wouldn’t pay for access yet if he didn’t get it for free through the AI Reform initiative. In part, that’s down to trust: “I don’t feel confident, even with Sana, to do confidential stuff,” he admitted, even if he has more trust in it than he does Silicon Valley-run AI systems. “Trust is a very big thing,” said Elf, who believes “being close to [users]” can help Sana enormously in contrast to farther-flung competitors.

Trust in Silicon Valley appears less of a concern in other countries: in May, the United Arab Emirates (UAE) became “the first country in the world to enable ChatGPT nationwide, giving people across the country the ability to access OpenAI's technology” as part of OpenAI’s Stargate UAE plans, a press release said. (UAE government officials did not respond to repeated requests for comment.)

Carissa Veliz, an AI ethicist at the University of Oxford, isn’t convinced that the investment for countries at a population-wide level isn’t worth it. “These systems don't come cheap,” she said. Veliz also worries about countries — particularly those outside the United States — locking themselves into dependence on US-based tech at a policy level. “It depends on how dependent we become on these companies that are very untrustworthy, and on a technology that seems very glitchy,” she said.

“When I advise governments, I advise them to make sure that they don't become dependent on a company or a cloud that they shouldn't be dependent on,” she said. “That they be as autonomous as possible.”

In Sweden, though, trust will be vital for its AI Reform initiative to succeed as the PC Reform did. “If I could really come to the level that I can feel this is secure enough that there will be no breach, and my stuff is my stuff, that would bring us to the next level,” Vare said.

While people like Vare will think twice and anonymize any personal content before feeding it into an AI, the principle behind these population-level experiments is simple. “The most powerful way [you can change things] today is really to just give it to the users and then let them just experiment with it,” said Elf.

“I think within a year, we will have seen at least two handfuls of countries that will have launched something similar,” she said, noting that organizations from Finland, the Netherlands and the UK had reached out to Sana. The UK and OpenAI reportedly discussed a potential plan in the past to give the entire UK population access to ChatGPT Plus, in what would have been a multi-billion-pound deal. The UK government told The Guardian, which broke the news, “We don’t recognize these claims.”

Whether you’re in Stockholm or Sharjah, it can feel like we’re all guinea pigs in an experiment adopting AI. For many, that’s deeply unsettling. But it can be equally exciting.

Peer out the window of Gilion’s offices, overlooking Birger Jarlsgatan in central Stockholm, and there’s a black billboard covering scaffolding in front of Norrsken House, a tech-focused co-working hub. On it, stark white text reads: “Prompt what matters. AI has given us unprecedented power. Let’s not waste it on another sales agent. Let’s use it to fix the problems that actually matter.” From his office a few hundred meters away, Werner said the AI Reform is “definitely a big bang moment, and you see it.” Whether the promise lives up to the billboard depends on how the next two million Swedes adopt AI.