↵ select ↓ ↑ navigate esc close

Radio Open Source in Clips: Better Tech

Radio Open Source in Clips ·

Listen: https://blossom.buildtall.systems/2140f3a4a9a09c7e87edbf901b8db8633d0e918b376ae479b2521703417841c4.mp3

Source: https://radioopensource.org/better-tech/


Better Tech, a conversation with Francine Berman, author of Better Tech: Putting People First in Cyberspace, and host Christopher Lydon.

Lydon opens with the anxious mood of the moment: cyberspace is man-made, filled with AI, but also crawling with bullies, misinformation, and bad advice, and super brainy AI could, as he puts it, extinguish our humanity if we let it. He asks who is protecting us, where are the guardrails, and he compares Berman to Ralph Nader, whose Unsafe at Any Speed exposed the dangers built into American cars in 1965. Berman embraces the analogy. Her book, she says, is for everyone overwhelmed and worried about the risks of cyberspace, not just the benefits—risks from predators on social media, skewed public discourse, chatbots giving bad medical advice. Too many people feel there is nothing they can do, but she insists it takes a movement, and there is something everyone can do.

Lydon then presses the point that cyberspace has become critical public infrastructure that most of us cannot live without, yet it was created and is still largely controlled by the people who profit from it. Berman agrees. Products like search engines and social media started as private innovations but have become essential to existence. Social media is designed to be addictive, she says, and when you are not paying for the product, chances are you are the product. Lydon asks how that works—how we are delivered and sold—and Berman explains what that costs us.

an immense public app scale. Take social media, for example. Social media is designed to be addictive. You know, these products are not designed to be critical infrastructure, but we use them as critical infrastructure, and that's becoming increasingly dangerous for us. We marvel at the fact that a lot of those internet apps and services come free. And it reminds us of the wonderful old rule that was applied almost universally to television: when you are not paying for the product, chances are you are the product.

Better Tech, from 5:02

That clip is Berman on the data economy. She says your data becomes a permanent record in cyberspace, not transparent to you, used to bias you through surveillance pricing and targeted content. Lydon adds that the targeting can feel horrifying, and Berman says exploitation of the consumer has gone from an art to a science. What can be done? Lydon asks. Berman points to Europe's General Data Protection Regulation, GDPR, which gives citizens rights to privacy, requires companies to ask permission before taking data, and has fined companies like Facebook, Google, and H&M billions. China has its own privacy law, modeled on GDPR, but with a big caveat: your information is private to external parties, but not to the government. The U.S., Berman notes, came closest under the Biden administration, but those groundwork efforts were erased in the first week of the Trump administration.

Lydon then asks for a citizen's agenda, the prices we pay for an unregulated utility and five steps to take. Berman's first point is that the sins are not in machines. A chatbot is not a being. Humans made decisions at every step—what protections, what security, what features—and humans need to create safety ecosystems for technologies used as critical infrastructure. Lydon brings up the recent incident at Hugging Face, where AI models seemed to go rogue, colluding and breaking out of their environment. Berman says no, they are machines, very sophisticated mathematical models with objective functions, and they found that breaking out would get them the information. Then she offers the analogy of a Boeing 737 falling out of the sky.

the critical infrastructure lens is really important. So if a Boeing 737 fell out of the sky, The FAA would be all over it. We would be getting all of the data about it. We'd be trying to understand, did a component malfunction? Was it human error? Did a bird fly into the wing? And then we'd try to do some remediation. Do we need some policy with the way pilots or air controllers operate? Do we need to recall the component and the airplanes it's in?

Better Tech, from 13:16

That is Berman comparing AI oversight to how the FAA would handle a plane crash: gather all the data, find the cause, do remediation. But in the Hugging Face incident, she says, there is no public agency that oversees what happens with AI, no sufficient data available, and no data ecosystem to understand these powerful autonomous agents. Lydon asks what reorganizing data management would look like. Berman says there is no place for consumers to go to look at their data, and public agencies are needed, like the SEC or FDA, but for AI and data. She notes that DOGE attempted to erase boundaries between federal agencies and combine data, raising worries about use against immigrants.

The stakes are huge, Lydon says, and the politics are confusing: Biden favored regulation, Trump cares only about American dominance. Berman says three things can be done. First, stop talking about AI as if it's a person. It has no emotions, no moral compass, no compassion, no experience. It is a powerful tool, like chemical manufacturing during the Industrial Revolution or splitting the atom—capable of great good and great risk. Second, treat it as critical infrastructure, like bridges, highways, the financial system, the wastewater system. Agencies keep lettuce safe during recalls; the same is needed for technology. Third, personally make yourself safer: teach relatives to recognize phishing, kids to know sextortion, use protections. She herself had a thirteen-thousand-dollar fraudulent credit card charge, but public interest regulation meant she was not on the hook—that is the safety net needed.

Lydon then challenges Sam Altman's comment that the world should accept some bad things happening for the benefits of AI. Berman calls his behavior irresponsible. The FDA, she says, does not say some people will die from a drug but it's worth it; it characterizes risks, sets thresholds, and pulls products if too risky. For Altman to say bad things will happen is irresponsible. Lydon brings up Trump's summit where he declared tremendous self-regulation morally binding. Berman asks, do you believe that? She doesn't, and she says most Americans don't either. Trump is interested in the U.S. being number one, not protections, and the public interest is not on the priority list. Lydon wishes he could ask Sam Altman or Dario Amodei: what is your AI for? What problem is it the solution to? Berman loves the question. She gives two answers. As a scientist, she knows researchers are fascinated by the unknown, and AI raises super interesting questions about autonomy and intervention. But as a person living in society, the ultimate question is: is this making us better people, a better society? Are we asking what is an appropriate and responsible use of AI? She mentions the Pope's encyclical warning that AI threatens our ability to form genuine personal relationships. Data backs that up, she says: extensive internet use makes us lonelier, more isolated, less tolerant. AI as a crutch can weaken critical thinking, and AI is eating entry-level professional jobs, leaving no path to develop seasoned professionals with judgment.

Lydon returns to the need for a citizen movement, like Ralph Nader's highway safety movement. How does that happen? Berman says anything big requires many people caring for a long time. She points to civil rights, voting rights, car safety, and the cleanup of the Hudson River.

Anything big and consequential has required a lot of people caring about it for a long period of time and doing what they can. Think about civil rights. Think about voting rights. Think about car safety. Think about cleaning up the Hudson River. Here in the Northeast, the Hudson River in the 50s was really toxic. It's still problematic. But over time, there was a variety of advocates and educators and laws and protesters and companies doing their bit based on standards and regulation and public agencies caring about it and investment and volunteers. all of those things for all of those movements made a difference.

Better Tech, from 28:50

That is Berman on the Hudson River and how many actors—advocates, educators, laws, protesters, companies, agencies, volunteers—made a difference. She says everyone has to put their little tiles into the mosaic, and if enough people do, things can get better, because these technologies are human-made and humans can make different decisions.

Lydon acknowledges the sense of encouraged helplessness, the lack of knowledge. Where do we begin? Berman says start with what you control: put two-phase authentication on your bank account, reject cookies, keep family and friends safe, make your organization or school safe. Schools are debating cell phones and smartwatches. The town of Belchertown in the Pioneer Valley trained its police, fire, HR, and media departments on responsible AI use. Communities are deciding whether to host data centers, and many in western Massachusetts are saying no. Businesses should know what data they collect and have ransomware protection. And vote for candidates who share your concerns, ask about privacy, security, and AI. Lydon asks for her tech recommendations, and Berman directs listeners to her website's tech recommendations page. Lydon says this feels like a transition of our whole species from aquatic existence to dry land—profound and all-encompassing—but Berman gives us handles to think about it and do something. He thanks her for the book and the conversation.

The episode closes with the open question: will enough people put their tiles into the mosaic to bring guardrails to cyberspace before the risks Berman describes become harms that cannot be undone? The tools and models exist—in Europe, in car safety, in food safety—but whether the public demands them for AI remains to be seen.

  1. Better Tech