Your Source for NPR News & Music
Play Live Radio
Next Up:
0:00
0:00
0:00 0:00
Available On Air Stations

This year in science: AI, James Webb Space Telescope research and climate change

ARI SHAPIRO, HOST:

It's time now for our science news roundup from our friends at NPR's Short Wave podcast, Regina Barber and Geoff Brumfiel. Good to have you here at the end of the year.

REGINA BARBER, BYLINE: Hey.

GEOFF BRUMFIEL, BYLINE: Hi there.

SHAPIRO: You usually talk us through three things happening in the science world each week, fresh off the journals and social media. But today, you're going to give us three big things in science from the year 2023. I couldn't be more excited about it.

BARBER: Awesome. Well, we're going to tick through the top three science threads we saw unraveling this year.

BRUMFIEL: We have all the scientific discoveries happening from the James Webb Space Telescope.

SHAPIRO: That's what that pile in your lap is.

BRUMFIEL: (Laughter).

BARBER: And fresh off of the COP28 conference, we'll talk climate.

BRUMFIEL: But we'll start with another really big theme, which I'm sure you're familiar with - artificial intelligence.

SHAPIRO: Yes. I'm actually not speaking right now. It is a...

BARBER: (Laughter).

SHAPIRO: No. It's not an AI replica of my voice, but it feels like, this year, everybody's afraid that AI is going to come for our jobs.

BARBER: Yep.

SHAPIRO: So what have you got for us?

BRUMFIEL: Yeah. I mean, you know, it's been a big worry for a lot of different fields. Visual artists have been sort of spooked by these image generators, like Lensa, Midjourney and Dall-E. Lensa's also being used to make AI social media avatars, which are, among other things, being used to push beauty standards to even higher, more unobtainable levels.

BARBER: And, of course, the big tech space program that has many a journalist, writer - you name it - in a panic was ChatGPT, the chatbot created by the company OpenAI. And over the course of the year, ChatGPT got more and more powerful and better at generating text. And it's not just ChatGPT. Like, many of the top tech companies all have their own AI, like Microsoft's Bing AI chatbot. That one made headlines this year for doing everything from professing its love for a journalist and trying to get him to leave his wife to, more recently, linking to conspiracy theories and lies when asked about elections.

SHAPIRO: Well, panic aside, a lot of the industries that have started to really integrate AI are in the STEM fields - science and tech - right?

BRUMFIEL: Yeah. So let's talk about medicine for a minute. You know, the AI revolution was already well underway in medicine before ChatGPT showed up. People have been developing algorithms to do things like diagnose diseases in scans and things like that. But with the language models, things are going even further. Some companies are floating systems to try and streamline medical notes and patient records. Others are rolling out programs that can generate correspondence between doctors and patients, and that's got some researchers nervous. I spoke to Marzyeh Ghassemi, who's at MIT and studies AI uses in health care. She cited one example where Microsoft just rolled out software to some hospitals that uses AI to write messages from doctors to their patients.

MARZYEH GHASSEMI: They're allowing it to draft this text as a suggestion of how a person should communicate with their patient, and that worries me.

BRUMFIEL: Because it really hasn't been tested, and we know AI can suffer from hallucinations - basically just make stuff up. It can give misleading information. And there's bias in the training sets that can discriminate sort of unconsciously against different groups of people. But the fact of the matter is, there's so much pressure on doctors in the medical system that these sorts of tools are going to get tried, and that's what's happening.

BARBER: Yeah. And in the sciences, AI is already starting to find its place, specifically in fields like chemistry and biology, where researchers have, like, a huge number of molecules and compounds to test. It can try to find ones that match the researchers' criteria, and people can synthesize the candidate chemicals or compounds in real life to see whether they work. And some labs are taking things a step further. Researchers at Lawrence Berkeley National Laboratory in California and Carnegie Mellon University in Pittsburgh have built robots that can do some of the synthesis and testing themselves, potentially further cutting down on time.

SHAPIRO: And, you know, as long as I've been alive, I've heard that STEM careers come with guaranteed jobs. What does it mean if AI can do advanced chemistry? I mean, are scientists going to be out of work?

BRUMFIEL: Yeah. You know, I've been thinking about jobs a lot this year as I've been reporting, and I keep coming up against this sort of basic truth - whether AI is being used in medicine, science or even things like surveillance and national security, it always seems like it comes back to the fact that it works best with a human in the loop. You know, I spoke to Sasha Luccioni, a researcher at an AI company called Hugging Face, and she put it this way.

SASHA LUCCIONI: I don't see generative AI models replacing people, but I can see them, you know, helping people or being used by people in their existing jobs in order to - whatever - to go faster or to be - you know, to have more creative ideas.

BRUMFIEL: And our colleagues at Planet Money really saw this firsthand. They used AI to make a series of episodes, and it worked. I mean, it did a lot of the work for them, but it worked best when they were giving it feedback. So I think that's something that - you know, we can hold out some hope that they're not just going to take our jobs - the bots.

SHAPIRO: OK, well, our second big topic is climate. What have you got for us?

BARBER: Yeah. So 2023 was a hot year - so hot that, once all the data is in, it's expected to be announced that this year was the hottest on record.

SHAPIRO: And I know scientists say, if we want to avoid the most catastrophic effects of climate change, humans need to keep global temperatures from increasing more than 1.5 degrees Celsius compared to preindustrial levels. How close are we to that number right now?

BRUMFIEL: The average temperature of the Earth over the last decade was about one degree Celsius higher than preindustrial temperatures, so we're definitely getting close.

BARBER: Yeah. And earlier this year, our colleague on the climate desk, Rebecca Hersher - she reported that to limit warming to 1.5 degrees Celsius, humans would have to slash greenhouse gas emissions more than 40% by 2030.

SHAPIRO: And we're not on track to do that.

BRUMFIEL: Yeah. Even the most ambitious plans to cut emissions wouldn't get to zero by 2050. So you're right - it's unlikely. But it's not like a cliff where humans are doomed as soon as we reach 1.5 degrees Celsius. We still have a lot of power to limit the negative impacts of climate change.

SHAPIRO: Well, right. A cliff implies that you either go over it or you don't. But if we don't hit 1.5, we have to keep it below two. And if we don't keep it below two, we have to keep it below two - like, it keeps getting worse, right?

BARBER: Yeah. I mean, that is stressful. But I do want to say it's not all doom and gloom. Like, humans are taking action around the globe. And NPR covered some of those efforts during our climate solutions week, like how Uruguay is using wind power and other green energy sources to help power their grid - 98% of the country's grid.

SHAPIRO: OK, we'll take hope from Uruguay. Let's leave this planet for a moment and check in on the James Webb Space Telescope, which continued sending back incredible images this year. Gina, you've reported on what this means for astronomy.

BARBER: Yeah. So this telescope has given astronomers a view into the early universe, like showing us the earliest galaxies, black holes, stars that we've ever seen. And what they look like is shocking to scientists because they are way more grown-up than scientists like astrophysicist Jorge Moreno expected.

JORGE MORENO: It's like if you went to a kindergarten and you saw a teenager.

BRUMFIEL: For perspective, I mean, galaxies were thought to form about a billion years after the Big Bang, you know, given the universe is around 13.7 billion years old in total. But now, JWST is really testing that hypothesis.

BARBER: Yeah. And also, to put these discoveries in perspective, another astrophysicist, Priya Natarajan, pointed out that we have already detected the oldest galaxy, the oldest black hole, just since JWST started its science operations last year. So she's pretty sure we'll discover more record-breakers in 2024.

SHAPIRO: Temperature records on climate, age records on black holes - it's a year of records. That's Regina Barber and Geoff Brumfiel from NPR's science podcast, Short Wave, where you can learn about new discoveries, everyday mysteries and the science behind the headlines. Thank you both.

BARBER: Thank you, Ari.

BRUMFIEL: Thanks.

(SOUNDBITE OF MUSIC) Transcript provided by NPR, Copyright NPR.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.

Geoff Brumfiel works as a senior editor and correspondent on NPR's science desk. His editing duties include science and space, while his reporting focuses on the intersection of science and national security.
Regina G. Barber
Regina G. Barber is Short Wave's Scientist in Residence. She contributes original reporting on STEM and guest hosts the show.
Related Stories