What happens when the tools you built start to outpace you? Astronomer J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz, reveals how artificial intelligence is transforming science at breathtaking speed — writing code, drafting research proposals, and uncovering exoplanets in decades-old data — pushing him to work “100 times” faster than ever before.
Subscribe To “Broadcast Retirement Network” On YouTube For Aging, Finance, Lifestyle, Privacy, Retirement, and Wellness programming Monday through Sunday at 7:30 AM ET.
Transcript:
Jeffrey Snyder, Broadcast Retirement Network
Professor, it’s so great to see you, sir. Thanks for joining us this morning.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
It’s a pleasure. Great to be on your show.
Jeffrey Snyder, Broadcast Retirement Network
And I appreciate you getting up a little bit early. I know you’re an early riser, but you’re on the West Coast of the United States, so I certainly appreciate you getting up a little early. You know, Professor, you wrote a great op-ed piece about artificial intelligence and its impact on science.
We’ll get to that in a few minutes. But I’m just wondering, because you play a big role in the scientific community, what makes a great scientist? How does someone become a scientist today, before artificial intelligence?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
I think the number one ingredient for that recipe is curiosity. And honestly, I think we all have it, at least at age five, if you’ve had a child or two or three, as I have. But you’ve heard the question, why, maybe a thousand times.
And that honestly is the heart of being a scientist. From there, yeah, it requires to be a professional scientist. There’s a lot of education involved, a lot of hard work.
But at the heart of it all, it’s, do you get excited by asking why and pursuing the answer to that question?
Jeffrey Snyder, Broadcast Retirement Network
So as you mentioned, it’s a rigorous process. I mean, I can’t, I’m not a scientist. I went to business school.
You can take that for what it is, but scientists go through undergrad, grad school. You may get your PhD. You may do postdoc work.
There’s a lot that, and you’re always testing your hypothesis. So, doctor, it’s pretty rigorous. And I’m wondering, in terms of today and your students, how do they kind of blend the use of that rigorous process of academia that, you know, your experience isn’t gained overnight or it’s not even gained in 12 months.
It’s gained over years and years. How do they kind of take that experience and blend it together in the educational process?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Slowly, which is, I think, the right way for we humans to learn. And I should say also by making just about every mistake possible. In the realm of astronomy, for example, when we, some of us are expert at collecting data with telescopes, you can’t show up at the world’s greatest telescope and collect data correctly the first time.
Can’t do it. I don’t care how many books you read, how many times you listen to me, no chance. And so we train them on smaller facilities like our Lick Observatory here in California so that they can cut their teeth, make those mistakes, and spend the time, and I mean nights, many nights, before they can become a really first-class astronomer on the greatest telescopes.
Jeffrey Snyder, Broadcast Retirement Network
So there’s a, what I like about science is that there’s a process. And we learned this in, I seem to recall learning it in like fourth or fifth grade, like the basics. I mean, obviously, I’m not at your level, but you come up with these.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Honestly, you learned it. I’m going to interrupt you. If I have the time, and I think I will, I will write a book called Why Science Isn’t Hard.
I’m announcing that in public for the first time today. You learned how to do science when you were three months old, when you were trying to stand up for the first time, and you fell. And you try it up again, and you fell.
And you then figured out gravity eventually so that you didn’t fall all the time.
Jeffrey Snyder, Broadcast Retirement Network
Okay, well, point taken, and I’ll stand corrected. But in terms of the, you know, I remember learning the, I guess, the practical process, you know, with the, in lessons in school. And you have a thesis, you propose a thesis, and you test the thesis.
And ultimately, through that testing, you either confirm the thesis, or you don’t confirm the thesis, or you have revisions to your thesis. So, you know, I’m wondering, in today’s environment with the use of artificial intelligence, how does that fit into, is it complementary to the scientific process that you and others use, or does it actually potentially replace the work that you are doing as a scientist?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Great question. It’s touching all aspects today. And each of us is at, I would say, at different stages in how much of the scientific method or scientific process we’re taking advantage of AI.
And some aren’t at all. Some are using it to code up, you know, write code. We’ve all heard that AI is amazing at writing professional level code.
It is. And so, we can use it in our science. Very much what we do is computer driven, and therefore software.
Some of us are using it to write proposals. Some of us are using it to brainstorm. I know colleagues who spend an hour a day chatting with one of the AI models to come up with their next ideas for science.
Some of them are writing their papers with it. It’s now impacting pretty much every aspect of the scientific process at the professional level, and it’s doing it very well.
Jeffrey Snyder, Broadcast Retirement Network
So, but is there still a role for the human scientist? I mean, you mentioned that kind of the back and forth writing the proposals. It would seem to me, just in my own experience with artificial intelligence, and I don’t want to put words in your mouth, but it can do some of the things that I do as a professional in financial services, but it doesn’t replace the relationships that I have with people.
It doesn’t really have the cognitive capabilities that I can to infer certain things. So, how about in the scientific community? Is it replacing scientists?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Not today, no. It’s accelerating many of us. I’m running at 100 times gas.
It’s crazy. But there is the potential. You run this, this is an exponential, Jeffrey, and we humans, we’re awful with exponentials.
Those of us that live through COVID now have a taste of it. We understand how powerful they can be. But still, as humans, we tend to be pretty, we tend to underestimate what an exponential can do.
AI has been growing exponentially for about five years. You run that clock for two or three years, what it can do today is going to be pretty amazing. Crazy what it’s going to be able to do tomorrow.
And in that level, replacement of humans, yes, is in play.
Jeffrey Snyder, Broadcast Retirement Network
But what, so is, you know, when you teach your students, and presumably, you know, you’re teaching undergrad and grad, and these are the future scientists of tomorrow, how do you teach them knowing that, one, they have to know the science. You have to have some basic knowledge as a person regarding the profession, but how do you teach them how to use AI knowing that, in some ways, it could supplant them as professionals? That makes sense.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Yeah, it does. Well, I’m going to, but you threw two questions at me. How am I teaching them to use AI today?
And let me answer that first. The skills that we need as humans today, again, this is moving so fast tomorrow, it’s hard to predict what we need. Today, we need to be able to read fast because it produces so much text so fast.
So reading is still at a premium. You can make it talk to you, so if you’re better at listening, fine. But still, you need to be able to comprehend on the spot or as quickly as you can and interact with the system.
In science, we need to be able to do what we call order of magnitude calculations. That’s, and colloquially, that’s a sniff test. When it produces a result, is it even close to correct?
Does it pass the sniff test? And that does require both knowledge of facts and in science, the ability to do what we call simple calculations, algebra. The third thing is to be able to visualize.
Can you draw down your ideas? So the chalkboard behind me, I torture students on all the time. They come in, they’ve got a great idea.
I go, put it on the board. And if they can, then they really have thought it through to an extent that they can communicate it. And if they haven’t, then we spend time learning how to communicate it.
Those are kind of three of the real key skills that I’m teaching my students so that they can interact successfully with AI today.
Jeffrey Snyder, Broadcast Retirement Network
Have they expressed any fear? From what I’ve heard and read, a lot of the graduates, now they may not have the post degrees, post-secondary degrees that maybe some of these higher level students have, but there’s some concern that, hey, I need to get a job and I don’t really have any material skills yet. I don’t have any experience.
Where’s that entry-level job? Have they expressed any concern to you and to their advisors about their future?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Yes, the grad students in our department, I think they do have those fears and should. And that’s led to at least some level of bifurcation between those that are embracing it and those that are abhorring it. I totally get both sides, honestly.
I’ll answer a bit more personally. I have three Gen Z sons, all who view AI probably a bit negatively or maybe more than a bit negatively.
Jeffrey Snyder, Broadcast Retirement Network
Interesting.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Some of the reasons you just enunciated. And so do I. I do not think this country is well-equipped for what’s coming because of what you said, because we are a country that insists on work for survival.
Jeffrey Snyder, Broadcast Retirement Network
And to that point, I mean, do you think any country is?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
I think there are many that are better off today for that precise issue than the United States.
Jeffrey Snyder, Broadcast Retirement Network
And so what could we do better in your mind in terms of preparing ourselves for better handling the exponential growth of artificial intelligence, sir?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
I don’t know much of anything about economics. Let me say that full front.
Jeffrey Snyder, Broadcast Retirement Network
Okay, but you have an opinion.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Well, I’m reading, let’s say it that way. It’s an informed opinion. But yeah, I’m reading statements like the one that came out from Bill Gates today.
I’m reading statements from my governor about UBI. I’m reading statements from CEOs of major companies talking about how the future will be without a workforce. I encourage, if people haven’t been reading that in that space, they should be, and thinking then about how they’re gonna vote going forward.
Jeffrey Snyder, Broadcast Retirement Network
So in terms of discoveries, let’s talk about, and again, you have an area of focus where it’s astronomy, astrophysics, oceanography. So those are pretty important disciplines. Let’s talk about, and this is an area that may not be aligned with that, your discipline, but how about like new drugs, treatments?
I mean, is AI positioned? And where does critical thinking come into play? Because I always like to think that, I grew up with the fax machine, before the fax machine, with a typewriter.
So I pride myself on my critical thinking. Where does that play in a role or does it?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
It’s essential. And I’ve had long conversations with my high school teacher in Rhode Island, Mr. Dimmick, who inspired me to become a scientist, about how challenging it is to engage humans in critical thinking. It is.
We’re good at it, but it’s hard work. And so actually getting your neighbor to think critically about anything, when they’d rather shoot the breeze, is difficult. So it plays a major role.
And I think it’s at the heart of where humans will go in the future. Will we absolve our critical thinking to AI and just let it do it? Or will we stay front and center in that process?
I don’t know.
Jeffrey Snyder, Broadcast Retirement Network
Yeah, I don’t think anyone knows. I think, if it’s clearly it’s evolving and it’s evolving faster than anything that we’ve ever experienced, maybe even a virus, although that tends to replicate pretty quick as well. But I don’t think we know where things are gonna go.
But I think we have to do our best, and maybe you would agree or disagree, we have to do our best to try to contain or put our arms around or adapt, because the future, it’s going to be the future.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Absolutely. And for me today, a terrifying one. As for drugs, I’m not a drug maker.
I know it’s having ginormous impact in genetics, in neuroscience, in pharmaceuticals. I know that from what I read. But yeah, certainly it’s not the science that I know well.
Jeffrey Snyder, Broadcast Retirement Network
Well, let me ask you, this is kind of tongue-in-cheek and off the beaten path, but there are hundreds of telescopes that have been collecting data for decades and decades. How can AI be used to find worlds that are beyond our solar system, maybe beyond our own galaxy to see if there’s life? What can it be used in the world of astronomy and astrophysics and oceanography to find?
I mean, what do you think you can uncover using artificial intelligence?
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
Yeah, the first round, which we would call, say, deep learning, the models that enabled people to identify cats and dogs with computers. Those first models actually were used on the Kepler dataset to discover exoplanets with the technique that the human experts were not able to do. So that’s eight years ago, seven years ago, I don’t know.
Today, the current systems are orders of magnitude better. I’m told now every time a new public dataset hits the internet, there are no fewer than 10 humans on the planet skimming through there with major AI systems to uncover whatever’s, the low-hanging fruit, the cream at the top. That’s gonna continue to happen for the rest of my lifetime.
And there will be major discoveries that otherwise humans would not have done, at least not at that speed. Full stop. So I guess- But data, I will finish by saying, data collection, the hundreds of telescopes, that still is a sticky point.
That is, AI cannot collect its own data today. That will change with robotics. But the expense, the labor required to collect data from the sky or to collect data across the ocean is expensive and human-led and will be at least for another, I don’t know, let’s say decade, I hope.
Jeffrey Snyder, Broadcast Retirement Network
Yeah, well, I hope for your sake and my sake and everyone else’s sake, we don’t want Skynet to replace us, I guess. Professor, we’re gonna have to leave it there. Really thought-provoking discussion.
I really appreciate your perspective. And look, we look forward to having you back on the program again very soon, sir.
J. Xavier Prochaska, Professor of Astronomy & Astrophysics at UC Santa Cruz
I appreciate it. Thanks, sir, for having me.
