Science Friday
Hosted by Ira Flatow
Unsurprisingly, artificial intelligence is still poised to be a major driver of technological change this year.
3,661 words · 6 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
Support for Science Friday comes from USA for UNHCR, the UN Refugee Agency. Here's a problem: How do you keep a house above freezing when the windows are gone and the power is out for 16 hours a day? In Ukraine this winter, the answer fits in a box: reflective insulation, plastic sheeting to cover broken windows, draft blockers for the doors. Simple materials installed the same day a team arrives. Your gift delivers them, and until December 31st, it's matched. Donate at unrefugees.org/friday.
WNYC is supported by Bear. Science is about asking questions, testing, trying again. It's not easy, but it's relentless. It evolves, and that's why science delivers. Learn more at sciencedelivers.com.
Science Friday is supported by Mathworks, creator of MATLAB and Simulink software for technical computing and model-based design. Mathworks: Accelerating the pace of discovery in engineering and science. Learn more at mathworks.com.
What advancements should we be looking out for in tech products this year, like AI? The concept here is that it's more conversational in style, hopefully more useful, easier to find the information that you want. It's Tuesday, January 7th. As always, today is Science Friday.
I'm SciFry producer Kathleen Davis. The tech world is always changing, and it can be hard to know what the important things are to keep an eye on. Lucky for us, each year the MIT Technology Review publishes a list of 10 breakthrough technologies. These are things that are poised to hit a tipping point and potentially change the way the world works. In the past, they've predicted the importance of things like wireless charging and reusable rockets, things that are being used a lot now. So what do tech journalists want us to keep an eye on this year? Here's Ira Flatow.
Joining me to talk through some of this year's entries is my guest, Amy Nordrum, Executive Editor of Operations at the fabled MIT Technology Review, based in Boston, Massachusetts. Welcome back to Science Friday.
Thanks, Ira. Great to be here.
All right. Before we get into the list, can you give us a peek behind the curtain? I mean, how are these breakthrough technologies chosen by your team?
Sure. It's a process that takes us several months every year. Everybody on staff pitches ideas and we're looking for technologies that are having what we call a breakthrough moment, which can mean a couple of different things. It could be for a new medical treatment, for example, that it's passed an important regulatory milestone, or perhaps it's a climate technology that's proven itself in a demonstration plant. So we define it in different ways, and that's part of the fun and the challenge of trying to put this list together. And in the end, we also want a range of technologies from different fields and areas, and you know, some stuff that maybe people have heard of and some other things that are going to be newer to them. And we put all these ideas together and we take a vote amongst our staff just to see what people are most interested in.
We have lots of debates and discussions about this list within our team. And eventually, we just have to make some tough calls and settle on the final list.
Are you looking mostly for a positive impact specifically?
Not necessarily. I do think this year, I would say, is a more optimistic list overall. But we also look for technologies that are just high impact in general, and that could be positive or negative. A couple of years ago, we had cheap military drones on the list.
Boy, was that right.
Yes, exactly. Those have, yeah, those have had a huge impact on shaping wars around the world. And we've, you know, there's other things that could go either way or have both sides. There's biometrics that we had a couple of years ago, obviously could make more life more convenient for us and hopefully more secure, but can also introduce new risks to privacy and personal data as well.
Okay, let's start with a topic that we've talked a lot about last year, and I'm talking about artificial intelligence. There are a few AI-related items on your list. Let's start with one of them, small language models. What the heck is that?
Well, you've probably heard about large language models. Those are AI models that are trained on all of the internet's data. And you can enter prompts or questions and they'll generate for you a written response or an image or a video of whatever you've described. And those have gotten lots of attention and many people are using them. But now there is growing interest in these small language models, which don't have as much training data and they don't have as many parameters, which are the values that get calculated during a model's training that it then uses to produce its results. But these small language models, even though they're smaller, they might be more suited to a specific purpose or task that someone's working on.
And some of these smaller models are doing just as well as the bigger ones on benchmark tests that the AI world uses to gauge model performance.
Wow. So, so what kind of tasks are we talking about here?
Well, for example, if you're a law firm that writes a lot of contracts, you don't necessarily need an AI model that's been trained on every Disney movie that's ever aired. You need one that knows contracts and is really well versed in that. Or if you're managing a factory, you need a model that can help you keep that operation running, maybe anticipating if there's maintenance that should be done to prevent breakdowns or troubleshoot a problem or keep track of records to show you what's coming in and out. So your model needs like a deep knowledge of your own business and the businesses that you work with. But it probably doesn't matter if it's read the entire Bible and complete works of Shakespeare in order to do that for you.
You know, we've, we've talked about how much juice, how much electricity these AI machines, these giant data centers use. Uh, maybe the smaller language models use less electricity?
Yes, it's not clear how much less really, but because of how they're made, they should be more efficient, so quicker to train and quicker, more efficient to run. And that should require less energy to do the things they need to do.
Yeah. Moving on, I don't know about you, but I saw dozens of advertisements for something called Google's Gemini over the holidays. And this has made it on to the breakthrough list, your breakthrough list. First of all, explain to me what this product is.
Yes, well, we have generative AI search on the list and Google's large language model called Gemini is powering this. It's specifically powering what's called AI overviews, which Google rolled out to the US in May. And what that means is you can type in a query and in Google's case, what's called an AI overview will show up at the top of the search results. I'm sure many of you have already seen this. This is a large language model summary of the topic or answer to your question. And it was generated just for you in the moment that you typed your query. It was not pre-programmed or scripted in advance. So the concept here is that it's more conversational in style, hopefully more useful to you, easier to find the information that you want. You don't have to go looking through a
ton of websites and a whole big list of links. The AI will do it for you. But these AI models, including Gemini, don't actually know if the overview that they're giving you is right or wrong. They're just assembling that answer based on their training, which is involved ingesting a lot of material from the internet and regurgitating some of that and predicting the next word in the sentence to kind of fill in the gaps.
So it's not just Google that is exploring this, I would imagine.
Right. Microsoft has infused Bing with generative search results as well, starting in September. That's powered by OpenAI's models. There's also, OpenAI now has hooked up ChatGPT into the live web, so it can go there to fetch answers and get updated information to any queries that you might have. And other companies like Perplexity, which has a chatbot interface of its own, are connecting theirs to the web. And in that case, they're relying on large language models by OpenAI and Anthropic as well to generate those results for you.
So now I understand this is quite controversial because it's a major problem for media industries that actually spend spend a lot of time and money collecting and researching stuff and then these guys trawl it all.
Yes, it is a bit of an existential crisis for the media industry, which we are both parts. That'll be an interesting challenge for us to try and solve. And there's other problems, too. As AIs, as these AI overviews were rolled out by Google, they the they started to give some nonsensical results. So there are some kind of funny examples where the AI overview suggested putting glue into pizza sauce to help the cheese stick to the pizza. And there was also a suggestion to eat rocks for nutrients that came up for one.
I'm writing this down now.
Don't do these things. Don't do these things at home. So, you know, Google has made some tweaks in response to those issues to not pull in as much user generated data or rely so much on sites that are known for satire and humor such as The Onion. But sometimes these also still just get things wrong. There was an example where they, one of the AI overviews said that Andrew Jackson, a former US president, graduated college in 2005, but he died in the mid 1800s. So these these new approaches to search will certainly cause challenges and changes in advertising and media. But there's also just some fundamental challenges around how much can we trust these and should, you know, which can we rely on them as well.
We have to take a break. And when we come back, more of the top tech breakthroughs to watch for this year, from robotaxis to stem cell therapies. Don't go away.
Science Friday is supported by PBS, presenting Going Wild. If you love learning about the natural world and meeting the scientists who help us understand it, be sure to check out the new season of Going Wild with Dr. Ray Wynn Grant. It's the chart topping and Webby Award winning podcast from the folks at Nature on PBS, now with a video companion series. In this season of Going Wild, host acclaimed wildlife ecologist Dr. Ray Wynn Grant. Heads out into the field to explore the animals, people, and stories that make nature worth saving. Stops include a hibernating black bear den in Arkansas and an underwater shipwreck site off the Florida coast. If you're looking for a place to start, check out the season's debut episode, where Ray joins beloved biologist and influencer Jordan
King in the Florida Everglades to learn about his unlikely journey into science and how to study a wetland when it's not very, well, wet. So don't miss out. Follow Going Wild with Dr. Ray Wingrand on Apple Podcasts, Spotify, YouTube, or wherever you get your podcasts.
I'm Kevin Ruse.
And I'm Casey Newton.
And together, we've been tech journalists for decades. Now, we're your hosts of Machine Gods, a new podcast from NPR that makes sense of what AI means for your work, your life, and all of our futures.
Twice a week, we'll be exploring the biggest moves in tech and give you unmatched access to the real titans of industry.
Listen to Machine Gods from NPR on the NPR app or wherever you get your podcasts.
Let's talk about another breakthrough on this list: robotaxis. I rode in a Waymo in San Francisco, where you can see driverless taxis, like Waymo's there all the time. But now they're coming to more cities, right?
Yes, that's right. We've heard about these for years, and companies like Waymo have been testing their services and collecting data and mapping out certain communities. But now they are really starting to expand into new cities and open up these services to the public. There are now autonomous taxis operating publicly in more than a dozen cities worldwide.
And that's what Elon Musk says also. He says his Teslas will be autonomous or a car version of it.
Yes, he just recently announced what he's calling the Cyber Cab and an autonomous bus as well, and has promised it for California and Texas this year. You know, he'll need to, his company will need to show that that's possible and that they've obtained their necessary regulatory approvals as well. But that, yeah, that was an announcement he made just a couple months ago.
Would you drive in a driverless car now? I mean, a lot of people are hesitant.
I also have taken a, yeah, a Waymo in San Francisco, and I I was quite impressed. I thought it was really interesting experience and I do think that more people will be trying these out for themselves and you know, perhaps feeling safer than they do themselves driving or you know riding with a a driver from another service. So we'll see. I mean, there's definitely going to be more availability of these services around the country and the world.
There are other companies, right?
Yeah, Waymo is definitely the US leader and the biggest in the US, but it's expanding to new cities. So so far it's been in LA, Phoenix and San Francisco, but now it's going to be in Austin and Atlanta. This year, there's also Baidu in China where they're operating many, many rides in Beijing and Wuhan in particular. And there's also here in the US, Amazon's project called Zooks, which is designed more like a bus. It doesn't have a steering wheel and it's launching to the public in Vegas this year and is also doing tests in San Francisco, Miami and Austin.
Wow. Wow. All right. Let's move on to some medical breakthroughs that made it on your list. Let's talk about an amazing new HIV medication that has tremendous results in trials. Let's hear more, please.
Yes, this is a drug that could be a much more convenient and accessible way for people to protect themselves against HIV. More than a million people are infected around the world with HIV every year still. This new drug is called Lenacapavir and it's administered as an injection once every six months. It's made by Gilead. And Gilead announced results in June showing 100% protection for more than 5,000 women and girls in Uganda and South Africa, and then more trial results in October showing 96% effectiveness at preventing HIV in more than 3,200 people of many other genders and sexual orientations. So this is a really exciting news and could definitely, you know, help us prevent HIV from spreading further.
Yeah, because the UN has a goal of ending AIDS by 2030. Could this medication make that possible?
It really could help if they are made widely available, although that is not a given. There are a couple other options out there for people already. There's preventative drugs known as PrEP, which can be pills or injections that have been around. You just have to take them every day, though, or ahead of time when you're exposed to a virus, which you might not always know, and people might not remember to take that medicine every day. So having just once every six months that you have to have this injection could make a real difference. But even once these medicines are approved, and this new one isn't yet approved for prevention, it's only approved for HIV treatment, there's still the matter of getting it out into the world, manufacturing it at scale and making it accessible at a price that people can actually afford.
All right, let's talk about another health breakthrough that, to me, following this for, I don't know how many years, always seems like it's five to 10 years away, and I'm talking about stem cell therapies. Are we closer now? How does stem cell therapies work and where are we?
We are getting closer, we think. So stem cell research has been ongoing for decades, as you said. And these are very powerful cells because they can be used to make any kind of tissue or cell in our bodies. And scientists have found ways to make these in the lab. You no longer need to use embryos to get them. You can make them from a bit of skin or whatever else. But it's been really hard to translate this research into proven medical treatments. And, you know, it's been a long time coming. But now there are two treatments in particular that we are highlighting on this year's list, one for type 1 diabetes and epilepsy, and they are both showing some real promise.
Wow. And how long until we see this therapy available, do you think? Tell me five years.
Yes, it's still a ways off from being widely available. So the trials so far have been very promising results. In some cases, you know, people that went under underwent the type 1 diabetes trial were able to stop taking insulin, and those that went through the trial for epilepsy had their number of seizures significantly reduced. But these are early trials, they're Quite small. The one for epilepsy by neuronal therapeutics only had fifteen patients in it. Overall, there's a couple thousand patients across a few dozen trials. And, you know, the vast majority of those trials are still still underway and will need to make their way through proper approvals to actually be available to patients.
Mm hmm. I know the holiday season is one of the biggest times for air travel. And aviation emissions are a large contributor to global warming, but there are companies working on cleaner jet fuel, right?
Yes, and that's really great because aviation and that whole sector is one area where it's been really tough to reduce emissions. And now, in addition to biofuels, which have been around for a while, there are some new ways of creating alternative jet fuels that are starting to scale up production. Initially, these fuels would be blended with petroleum to reduce emissions in flights, but eventually you could have planes that run on one hundred percent of these fuels. And generally speaking, these fuels can be used in planes as they are. There are no modifications required to the plane.
Okay, let's end with a breakthrough in an industry we don't always think about as having technological advancement, and I'm talking about construction. And in this case, tell me about this steel plant in Sweden that will release almost no carbon emissions. Isn't isn't steel made with carbon?
Yes, it requires a lot currently. This plant is the first of its kind or will be the first of its kind in the world. And steel is mainly an alloy of iron and carbon, and it's made generally by putting iron ore into a series of furnaces with some coke, which comes from coal and limestone to remove impurities to create a crude form of iron, which is the key ingredient in steel. But now companies are green steel companies like Steggra, this one in Sweden, are working on making steel by using hydrogen produced from renewable sources to convert the iron ore into iron, which is the dirtiest step of the steelmaking process. And this company called Steggra is
raised almost seven billion dollars and they're building the first industrial sized plant using this method anywhere in the world, and they're hoping to start producing essentially low or zero emissions steel there in twenty twenty six.
So you're saying they don't have to use carbon anymore to make their steel?
So they're using hydrogen produced through electrolysis. They're using electricity produced from wind and hydropower to split water. And then they're using that hydrogen gas to pull oxygen out of the iron ore to make the metal that we know of as iron. And they're not the only ones doing this. There are some other major steelmakers that are investing into similar technologies. And there's also one that's trying to build another plant in Sweden to do the same thing. So it's exciting to see this progress. They still need to show this all works at scale and obviously get buyers who will pay a little bit more of a premium for this steel. They have locked in a couple of deals in that direction. But hopefully if it works and if it catches on, this could really help us bring down emissions from the steelmaking process.
Yeah, well, thank you for taking time to be with us today.
Amy Nordrum, Executive Editor of Operations at the MIT Technology Review, based in Boston, Massachusetts. And if you want to read the full list of breakthrough technologies, go to technologyreview.com.
And that's it for today. Tomorrow, we're bringing you a mind bender. What would happen if scientists created artificial bacteria or other organisms that were a mere image of normal life? Why some scientists are concerned. Lots of folks help make the show happen, including
Sandy Roberts. Robin Casmer. Jordan Smudjic. Charles Bergquist.
And many more. I'm Sci-Fry producer Kathleen Davis. Thanks for listening.
Page updated November 5, 2026 · transcript and summary are machine-generated