AMY GOODMAN: This is Democracy Now! I’m Amy Goodman, with Juan González.

We end today’s show looking at how the Chinese startup DeepSeek, which uses a free, open-source artificial intelligence model, has upended Silicon Valley. It’s new, low-cost AI tool R1 has surged to the top of Apple’s App store, overtaking OpenAI’s ChatGPT. Meanwhile, artificial intelligence giant Nvidia lost almost $600 billion in value Tuesday, the biggest single-day loss for a public company. In December, DeepSeek said its model only took two months and less than $6 million to build, despite U.S. restrictions on chip exports to China — a tiny fraction of the cost that U.S. tech giants have spent on AI. Just last week, President Trump announced Stargate, a $500 billion project, to boost AI infrastructure in the U.S., and he promised it would create new jobs.

For more, we’re joined from Los Altos, California, by Ramesh Srinivasan, professor of information studies at UCLA, host of the podcast Utopias, author of Beyond the Valley: How Innovators Around the World Are Overcoming Inequality and Creating the Technologies of Tomorrow.

In this last few minutes we have, Professor Srinivasan, can you talk about the significance of DeepSeek?

RAMESH SRINIVASAN: Yes. I mean, it’s very, very profound. I mean, basically, for $6 million, they were able to develop a large language model, which is basically the brains for future AI systems of all kinds, for $6 million, compared to hundreds of billions of dollars that were allocated for initiatives like the Stargate project. Now, this is not apples to apples. The $6 million is not totally equivalent to the $500 billion and the investments tied to that.

But the key issue is this: DeepSeek was able to train and refine its models using open-source kinds of content, getting input from communities of developers all around the world. And this is very insightful. It’s an example of something that we’ve always known in tech: that innovation is not just creating the latest iPhone 15 versus 16 or just pouring tons of money into energy and other expenses in this sort of, you know, resource-unconscious way, but it’s actually being resourceful. And the open-source community is why DeepSeek was able to basically perform very close to the level, if not stronger, than ChatGPT’s latest, or at least previous to latest versions, for a fraction of the cost. And this is a key, key breakthrough, and this is why we’re seeing so much volatility in Silicon Valley as we speak.

JUAN GONZÁLEZ: And why is this being seen as a threat rather than as enormous progress in terms of development of artificial intelligence?

RAMESH SRINIVASAN: Right. It’s very much a threat, because many Silicon Valley models are based on proprietary sorts of platforms or proprietary data. They are gathering data from all of us to inform their AI systems, which are going to be used primarily for their benefits, never mind that there are various fairly unclear and possibly empty promises about jobs that these AI infrastructures, like Stargate, will create. So, this is just a remarkable example of one thing that’s always been powerful about the internet and the web, which is people can support one another and build on one another’s work. And that’s what DeepSeek has actually shown us. And they’ve validated that in various reports, and a number of researchers I’ve looked at have validated their findings. But the Silicon Valley model, which has in its past and even to this day sometimes uses open-source input, has primarily been gated communities, you know, walled gardens, algorithmic extremist content, as we’ve talked about many times before.

JUAN GONZÁLEZ: Yeah, we only — we have less than a minute, but I wanted to ask you — on Tuesday, OpenAI announced a new product called ChatGPT Gov, built exclusively for the —

RAMESH SRINIVASAN: Right.

JUAN GONZÁLEZ: — United States government. Can you talk about that briefly?

RAMESH SRINIVASAN: Yeah, absolutely. Well, look, there’s more and more of just a very direct and explicit marriage that we’re seeing between the Trump administration and tech oligarchs. I mean, we saw it at the inauguration. We’re seeing it in this initial week and weeks of campaign promises. So it’s really no surprise that Big Tech is kind of colliding with the state, because it’s all about the concentration of power. But I want to remind everybody that even back in the Obama era, Google was the biggest lobbyist during the Obama administration. So we need to be vigilant and ensure that AI systems and technologies of all kinds support laborers, citizens and people around the planet. And that’s what’s woefully missing in most discussions of DeepSeek, OpenAI and Big Tech, in general.

AMY GOODMAN: And finally, in 10 seconds, how does this relate to TikTok, if it does in any way, with the decision coming down on whether it will be banned?

RAMESH SRINIVASAN: Very important question. These are all data grabs, right? So, part of Musk’s ownership of Twitter/X is the data that can be grabbed to form his own AI initiatives which are trying to compete with OpenAI. It’s going to be a very similar issue when it comes to TikTok. The acquisition of TikTok is an acquisition of a largesse of data, at least American data. America has the largest number of TikTok users in the world. So, that data can all be mined to reconstruct these types of chatbots, which, again, are the brains of different types of consumer-facing AI systems.

AMY GOODMAN: We have to leave it there, Professor Ramesh Srinivasan —

RAMESH SRINIVASAN: Thank you.

AMY GOODMAN: — of UCLA. I’m Amy Goodman, with Juan González. Thanks for joining us.