StarTalk Radio - Your Brain on ChatGPT with Nataliya Kosmyna

Episode Date: September 19, 2025

What happens to your brain when you use AI? Neil deGrasse Tyson, Chuck Nice, and Gary O’Reilly explore current research into how large language models affect our cognition, memory, and learning with Nataliya Kosmyna, research scientist at the MIT Media Lab. Is AI good for us? NOTE: StarTalk+ Patrons can listen to this entire episode commercial-free here: https://startalkmedia.com/show/your-brain-on-chatgpt-with-nataliya-kosmyna/Thanks to our Patrons Jacqueline Scripps, Jose Mireles, Eric Divelbiss, francisco carbajal medina, Sahil Pethe, Vivekanandhan Viswanathan, Kurt R, Daniel D. Chisebwe, Landslide, Sebastian Davalos, Bob Case, Mark Rempel, Lucas Fowler, Cindy, Wizulus Redikulus, Hector Alvarado, Matt Cochrane, Ari Warren, Mark, Jorge Ochoa, Leena Z, Donald BeLow, Zach Woodbury, Jeffery Hicks, Ibolinger, Subri Kovilmadam, Danielle Stepien, Justin Akins, Richard, Tai Vokins, Dan O'Connell, Evelyn Lhea, Siva Sankar, Jack Bremner, mcb_2011, Saronitegang, dante wisch, Adnrea Salgado Corres, Jarrod C., Micheal Maiman, Ivan Arsov, Patrick Spillane, Aarush, Brad Lester, Anna Wolosiak-Tomaszewska, Jon A, Ali Shahid, K. Rich Jr., Kevin Wade, Suzy Stroud, Expery Mental, Ian jenkins, Tim Baldwin, John Billesdon, Hugo, Mason Lake, Judith Grimes, G Mysore, Mark Stueve, Cuntess Bashory, Jock Hoath, Payton Noel, and Leon Rivera for supporting us this week. Subscribe to SiriusXM Podcasts+ to listen to new episodes of StarTalk Radio ad-free and a whole week early.Start a free trial now on Apple Podcasts or by visiting siriusxm.com/podcastsplus. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Chuck, if the forces of AI are not big enough in society and our culture, we're not got to think about what AI's effect is on our brain? I'm going to say that there is no help from my brain, so it does not make a difference. I know, but Neil, if you lean into these large language models and it takes away some of our core skills, surely that can't be an upside to that. Once again, Gary, not going to affect me at all. Coming up, StarTalk Special Edition, your brain on AI. Welcome to StarTalk, your place in the universe where science and pop culture collide.

StarTalk begins right now. This is StarTalk, special edition. Neil deGrasse Tyson, you're a personal astrophysicist, and when it's special edition, you know that means we have Gary O'Reilly in the house, Gary. Hi, Neil. All right. We got another one of these. We're going to connect the viewer, listener, to the human condition.

Yes. Oh, my gosh. But let me get my other co-host introduced here. That would be Chuck Nice. Chuck, how are you doing, man? Hey, man. Yeah, that when you know it's Chuck, it means it's not special at all. Oh.

But we've got you because you have a level of science literacy that, Oh, my gosh. You find humor where the rest of us would have walked right by it. And, you know, that's part of our recipe here. That's very cool. Yeah, I appreciate that. Yeah, so, Gary, the title today is AI good for us? Okay, well, here's the answer.

No, let's all go home. No, okay, that's the end of the show. Let's all go home, people. Well, this was quicker than I expected. This was very quick. I mean, yeah, you know. So, Gary, what have you set up for the day? Well, Lane Unsworth, our producer over in the LA office and myself, we sort of noodled.

And this is a question that's been bouncing around a lot of people's thought processes for a while. So all over the world, people are using LLM's large language models for their work, their homework, and plenty more. Besides discussions of academic dishonesty and the quality of work, has anybody actually taken the time? to stop and think about what this is doing to our brains. Today, we are going to look at some of the current, and I really do mean current, time and space, this moment, researched into the impact of using an AI tool can have on your cognitive load and your neural and behavioral consequences that come with it. And the question will be, does AI have the opportunity to make us smarter or not? I like the way you phrased that, Gary. It was a very diplomatic.

I know. Smarter or not? Or not? And does it have the opportunity to do so? Okay. Smarter or dumber? That's what you mean. I didn't say those words. Well, here on StarTalk, we lean academic when we find our experts. And today is no exception to that. We have with us, Natalia Cosmina, dialing in from MIT. Natalia, welcome to Start Talk. Thanks for having me excited to be here with you. Excellent. You're a research scientist at the one and only MIT Media Lab. Oh, my gosh. If I had like another life in a career, I would totally be on the door crepts there wanting to get a job. And if I had another life and career, it wouldn't exist. I'd shut it down immediately because let's be honest, science is a hoax. some people do want you to believe that you know it's like science has 99 problems and virality ain't one right

right there you go and you're in the fluid interfaces group you are trained in non-invasive brain computer interfaces bc i'm guessing that means you put electrodes on the skull instead of inside the skull, but we'll get to that in a minute. And you're a BCI developer and designer whose solutions have found their way into low Earth orbit and on the moon. We want to get into that. So let's begin by characterizing this segment

as your brain on chat GPT. Let's just start off with that. What a great topic, Neil. Is there any way I can help you with that? So, so, You research what happens when students use chat GPT for their homework and for their, what have you found in these studies? Yeah, so we run a study that's exactly the title, right? Your brain on chat GPT, accumulation of cognitive debt when using an AI assistant for essay writing tasks.

So we did a very specific task that we're going to be talking right now about, which is essay writing. We invited 50 students from Greater Boston area here to come in person to the lab and we effectively put those headsets you just mentioned on their heads to measure their brain activity when they're writing an essay. And we divided them in three groups. We asked one group, as you might already guess where that's heading, to just use chat GPT.

That's why paper is called your brain on charge GPT. It's not because we are really, really singly out chat GPT. It's just because we use chat GPT in the paper. so it's pure scientific. So we asked one group of students to use only chat GPT to write those essays. Another group to use Google, so search engine, to write those essays. And the third group to use their brain only. So no tools were allowed.

And we give them topics which are what we consider a high level, right? For example, what is happiness? Is their perfect society? Should you think before you talk? And we give them a very limited time, like 20 minutes to write those essays. And we finally, of course, looked into the outputs of those essays, right? So what they actually written? How they used CHATGPT?

How they used Google? And, of course, we asked them a couple of questions, like, can they give a quote? Can they tell us why they wrote this essay and what they wrote about? And then there was one more final fourth session in this study, where we swapped the groups. So students who were originally in CHRGPT group, we actually took away the access for this fourth session. and vice versa was true. So if you were, for example, you were not our participant,

but if you were ever to come to Cambridge and be our participant, and let's say if you were actually... I'm not putting anything on my head. I'm just letting you know right now. Okay. Come on, it's the future. It's the future.

Now, the problem is he'd have to take off his tinfoil hat when he got there. Yep, yep. I see that happening. Regardless. So if you were, for example, in our participant in Brain Only, group. We actually for this first session would give you access to your GPT. And again, we measured

exact same things, brain activity, what actually was an output, and asked couple questions. And what we found are actually significant differences between those three groups. So first of all, if you talk about the brain, right, we measured what is called brain functional connectivity. So let's, in a layperson terms, like I'm thinking here having three of you talking to each other talking to myself. So that's what we measured. Who is talking to who am I talking to Neil or is Neil talking to you? So directionality. So who talks to who in the brain? And then how much talking is happening? Is it just, hi, hello, my name is Natalia or actually a lot of talking. So a lot of flow of data is being exchanged. So that's literally what we actually measured.

you found significant difference and then some of those are ultimately not surprising you can think logically if you do not have any let's say you need to do this episode right now right and i'm going to take away all your notes right now all of the external help and then i'm going to measure your brain activity how do you think it's going to turn out you're going to have like really your brain on fire so to say because you need like okay what was her name again what was the study what what is happening right you need to really push through with your brain, like you have memory activation, you need to have some structure, like, and now you don't have notes for the structure of this episode, right? So you need like, what was the structure? What we did this? What we are talking about?

What is, you know, you really have nothing to fall onto. So, of course, you have this functional connectivity that is significantly higher for brain-owner group compared to the two other groups. Then we take search engine group, Google. And actually, there's just as a prior research, There's a ton of people about Google already. We actually, as a humanity, right, we are excellent in creating different tools and then measuring the impact of those tools on our brain. So there's quite a few of papers we are citing in our paper. For example, there is a paper, spoiler alert, called Your Brain on Google from 2008.

Literally, that's the name of the paper. So we've actually found something very similar to what they found. There would be a lot of activations in the back of your head. This is called visual cortex or occipital cortex. It's basically where a lot of visual information processing. So right now, for example, someone who's listening to us and maybe they are doing some working parallel, they would maybe have some different tabs open, right?

They would have, like, one is like YouTube tab, and others they would have like some other things that they're doing. So, you know, you're basically jumping between the tabs, looking at some information, maybe looking at the paper while listening to us. So this is what we actually seen, and there's a plenty of papers already showing the same effect. But then for the LLM group, for CharGPT group, we saw the list of these functional connectivity activations. And it doesn't, again, means that you became dumb or you

Yes, it does. There's actually quite a few papers specifically having in the title laziness, and we can talk about this with other results, but from brain perspective from our results, it doesn't show that. What it actually shows that, hey, you have been really exposed to one very limited tool, right? You know, there's not a lot of visual stuff happening. Brain doesn't really struggle when you actually use this tool. So you have much less of this functional connectivity. So that's what we found.

But what is, I think, interesting and effective, maybe heading back to this point of laziness, and some of these maybe a bit more, I would say, nefarious results are, of course, other results that are relevant to the outputs, to the answers. So first of all, what we found, that the essays were very homogenous. So the vocabulary that was used was very, very similar for the LLM group. It was not the case for the search engine and so for the brain-only group.

I'm going to give you an example. And, of course, in the paper, we have multiple examples. I'm going to give you only one. Topic happiness. So we have LLM, so chat GPT users, mentioning heavily the words career and career choice. And surprise, surprise, these are students, I literally just mentioned this. Of course, they're going to more likely talk about career and career choices. And again, who are we ultimately to judge what makes the person happy, right?

No, of course. But don't forget, the two other groups, they are from the same categories. They are students in the same geographic area, right? However, for them, these words were completely different. For the Google, for the search engine, students actually heavily use vocabulary giving. and giving us, and then brain-only group was using vocabulary related to happiness and true happiness. And this is just one of the examples. And then finally, to highlight one more result is responses from the participants themselves, from those students.

So we asked literally 60 seconds after they gave us their essays. Can you give us a quote? Any quote, any length of the quote, of what you had just written. It can be short, long, anywhere in your essay, anything. 83% of participants from LLM, from Chi-GPT group could have not quoted anything. That was not the case for brain and search engine groups.

Of course, in sessions two and three and four, they improved because surprise, surprise, they knew what the questions would be, but the trend remained the same. It was harder for them to quote. But I think the most ultimately dangerous result, if I can use this term, though it's not really scientific, but something that I think a lot of inquiry actually is required to really look further into this, it's almost on philosophical, I guess, level is ownership question. So we did ask them if how much of percentage of ownership do they feel towards those essays?

And 15% of chat GPT users told us that they do not feel any ownership. And of course, a lot of people, especially online, mentioned, well, they haven't really. reading this essay, of course you didn't feel any ownership. But I think that's where it actually gets really tricky, because if you do not feel that it's yours, but you just rocked on it, does this mean that you do not care? We don't obviously push it that far in the paper, but I think this is something that definitely might require much further investigation. Because if you don't care, you don't remember the output, you don't care about the output, then what ultimately is it for? Why are we in here, right?

Of course, it's not all dark gloom and everything is awful, right, and disastrous. I mentioned that there's this fourth session. Not everyone came back for this session, so actually sample size is even smaller for this. Only 18 participants came back. But what we found is that those who were chat GPT users originally and then lost access to chat GPT, their brain connectivity was significantly lower than that of the brain-only group. However, those who were originally brain-only. group and then gained access to CHAPT, their brain connectivity was significantly higher than

that of the brain-only group. What it could potentially, and I'm saying potentially because, again, much more studies would be required, means that timing might be essential. Basically, if you make your brain work, well, and then you gained access to the tools, that could be beneficial. But of course, it doesn't mean that it's one second of work of the brain and then you use the tool, right? something like let's say you're in a school and maybe first semester you have you learn your base

of whatever subject it is without any tools like old school way and then on the second semester you didn't become an expert right in one semester for school year but you at least have some base and then let's say in the second semester you gained access to the tool right so it might prove actually beneficial but again all of this is to be still shown and proven we literally have very few data points. But the tool is now being really pushed on us everywhere. So you could be affecting best practice for decades to come based on what a teacher might choose to allow in classroom and not. So what are you measuring? You know, you put the helmet on. Are you measuring a blood flow to, is it, neuro-electrical fields? In our case, we measure,

measuring electrical activities, or there's multiple ways of measuring things? Is that the EE? E.E.G. Yeah, electroencephalography, yes. Right. Okay, so that just tells you, and since we already know in advance,

what parts of the brain are responsible for what kinds of physiological awareness, right? And if you see one part of the brain light up versus another or no part light up, that tells you that not much is happening there, Is that a fair? Yeah, it's a bit simplified, but kind of fair way. And it doesn't mean that it's very important.

It's not that that part didn't, doesn't work, right? Or like it atrophied itself like we saw in some mental pressure. No, no, no. It just means you started as a dumbass and you still are one. It's dumbed. Well, what happened? This guy's brain just went completely dark. It doesn't go dark.

Like, listen, I'm going to give you one example, right? It's like back to this crazy example. above 3% of our brain versus 100%. Like, if you were to use not 100% of your brain, like literally, we would not have this kind of section right now at all. So it's very important to understand. We use our brain as a whole. Of course.

No, we're not. We are way past that. We're not in that camp. That was just a joke. We understand that your brain is constantly working. A lot of it actually. just to run your body. So, you know.

Takes up a lot of energy. Takes up a lot of energy. But back to the energy, and I think this is like super important, it still takes much less energies and even, you know, 10 requests from charge GPT or from Google. And this is beautiful because our body, right, so imperfect as a lot of people call it in our brain, so improper. Which is very old, ancient, as some people say, computer,

still is the most efficient of machines that we all have, right? And we should not forget that. People and all of the AI labs right now around the world try to mimic the brand. They try to pull so hard all of those preprints that you see in and archives, the service that hosts those papers. How can it be similar? Can we ensure that this is similar, right?

And so there is something to it because we are actually very efficient. But we are efficient almost to the limit of the shortcuts that actually makes in a lot of cases a bit too efficient, right? Think about, like, hey, you really want to look for these shortcuts, so make things the easiest. The whole goal of your brain is to keep you alive, not to use charge, GPT, or LLM, not to do anything. No, the only ultimate goal, let's keep this body alive. And then everything else adds on, right? And so this is how we are running around here. We are trying to obviously then figure out how we can make life of this body as easy as we can.

So, of course, these shortcuts are now, as you can see, used in a lot of social media, which obviously heavily talked about. And we know about some of those dark patterns, as they are known, are heavily used. And some of them are designed by neuroscientists, unfortunately, because it feeds back into the needs of the brain, constant affirmation, fear of missing out. All of those are our original, original design by the nature, right? Phenomena. And of course, now we can see that LLMs would be and are getting designed by those as well. Wait, Natalia, just a quick insert here. So I had not thought to compare, just as you described,

the energy consumption of an LLM request in chat GPT and the energy consumption of the human brain who achieve the same task, for example. Are you factoring in that I can say, write me a thousand word essay on Etruscan pottery, okay? And 30 seconds later, here it comes. And you can go to the servers or whatever, or the CPU, look at how much energy that consumed. Meanwhile, I don't know anything about Etruscan earns. So I will go to the library and I'll go and it'll take me a week.

Can you add up all the energy I did expend over that week thinking about it and then compare it to the chat GPT? Do they rival each other at that point? So, definitely, that's an excellent point, right? So theoretically to answer your question, we can, right? The difficulty, actually, would be on the LLM part, not on our part, because we do not have, you know, there's a lot of these reports, right, in their LLM consumption per all of these procents for the prompts, right? But what a lot of companies, well, actually no, almost no companies are releasing, is what it took for training, right?

So for you, it took 30 seconds of thinking, and I hate, hate, hate this word thinking when we use it for LLMs, right? That's not thinking, right? But like, let's keep it for now thinking. That's what you see on the screen. But ultimately, you do not know. And neither you know myself, there's no public information how long it took for you to be trained to actually give you some pottery.

Most likely, my, my assumption, this is obviously subjective. I do not have data, so I need to be very clear here. But my estimate from overall knowledge that is available, you're going for a week to the library, not going to be more beneficial for your brain because you will talk to other people, getting this chat of the library,

and all of the process information. Your brain will struggle. Your brain actually does need struggle. Even if you don't like it, it actually needs it. you will learn some random cool things in peril, maybe excluding pottery, and that will still take less for your whole body to work, right, than actually that's 30 seconds of the pottery from a chart GPT. Again, very important here as a note, we do not have the data from LLM perspective, so this is just my subjective one.

I'm Joel Cherico, and I support StarTalk on Patreon. This is StarTalk with Neil deGrasse Tyson. So Natalia, you've obviously chosen essay writing for a reason. It is a real, it is a challenge on a number of levels. Your research is fresh out the oven. It's June 2025, and we're only a couple of months down. a road from there as we speak right now. Have you explained to us cognitive load and then cognitive load theory and how it blends in and how it sits with your research? Please. Absolutely. So

just to simplify, right? So what actually happens is for there are different types of cognitive load. Actually in the paper we have a whole small section of this. So if someone actually wants to dive into that, that would be great. There are different types of cognitive load. And the whole idea is that it's how much of the effort, right, you would need to be on the task or to process information in the current task. For example, if I'm going to stop right now talking as I'm talking, I'm going to start just giving you very heavy definitions. Even if you're definitely interested in those, it will be just harder for you to process. And if I were to put this brain sensing device on you, right, the EEG cap that I mentioned, we would definitely see that spike because you would try to follow. And then you'll be like, oh, it's interesting, but really gets hard and hard if I'm going

to just throw a ton of terminology on you, right? So that's basically what, and this is just simplification, right? There's definitely, like, check the paper, and there's so, so much into that. The idea for the cognitive load and the brain, though, is that all it is studied before us, so not in our paper, we just talk about this, but there are multiple papers and some of them beside in our paper, is that your brain actually, in learning, specifically, specifically in learning, but also in other use cases, but we are talking right now, learning, actually needs cognitive load.

Like, you cannot just deliver information on this, like, platter. Like, here you go, here's information. There are studies already pre-LLM, so pre-large language models use pre-chartboards that do talk to you about the fact that if you just give information as is, a person will get bored real fast. And they'll be like, yeah, okay, whatever. there will be less memory, less recall, less of all of these things. But if you actually struggle for the information on a specific level, right, it should not be very, very hard. So if you are

cognitively overloaded, that's also not super good because basically you can give up, right? It's actually a very beautiful study. From 2011, I believe, it's actually measuring pupil dilation. So literally how much pupil dilates when you are giving very hard to understand words. words and vocabulary, and you literally can see how when the words becoming longer and harder, you basically, it kind of shuts down. Like, it's like giving up. Like, I'm done here, processing all of that. I'm just going to give up, right? So you don't want to get a student or someone who's learning something new on this give up. Information is already delivered to you within 30 seconds or three seconds or 10 seconds, and you haven't really struggled. There is not a lot of this

cognitive load, and a lot of people would be, but that's awesome, right? That's kind of promise of these other lamps and a lot of these tools but we do not want to make it too simple right we do not want to take away this cognitive lord and it sounds like almost it sounds like cognitive lord don't we want to take it away no you actually do not want to take it away what you're describing right now is the uh basis for all video game design yes that's what you're describing right now what they want to do is make it just challenging enough. If it's too challenging, you give up on the game. But if it's too easy, you also give up on the game. But if it's just challenging enough so that you can move to the next level

and then struggle a little and then overcome the struggle, they can keep you playing the game for very long periods of time. And so it's a pretty interesting thing that you're talking about. But what I'm interested in beyond that is when you talked about the cognitive load, I'm thinking about working memory. Yeah. But then I'm also thinking about the long-term information that's downloaded within me. So let's say I'm a doctor, right? And it's just like, oh, he's suffering mild dyspania because of an occlusion in the right coronary, blah, blah, blah, blah, blah, blah, blah. For a doctor, that's a lot of information, but they're so familiar with the information, it's not a stress on their working memory.

So how does that play into, in other words, how familiar I am with the information already and like how well I can process information naturally, how does that play into it? And Chuck, did you just describe your own condition? I don't know what you said, but you were way too fluent at it. Yeah, he would like doctor house. Oh, my God. He knew it. Neil, you are too damn funny. But guess what?

You're right. How about that diagnosis? By the way, I could have kept going. That was only one problem that would happen. But go ahead. It's actually perfect, right? It was a perfect example right now in this conversation between Chalka and Neil because Neil is like, I have no idea what you just said.

Maybe it's a nonsense. Maybe it's actual real stuff. It's perfect. If you have no idea, so you are basically novice, right? So you have no base. You can really be like, what is happening? You will have confusion. You will have heightened cognitive lord, right? You would be like, have I heard of anything like that before? So you will try to actually try to do a recall. Like, okay, I haven't heard it. I, it's not my error expertise. What is happening here? And obviously, you will now, because you heard all of these words that you have no idea about, and if the topic is of the interest to you overall, you will try to pay attention, make sense

out of it, maybe ask questions, et cetera. But if you are effectively trained on it, right, so you're a doctor, you are a teacher, you are an expert in the area, we see that there are significant differences. Well, first of all, because you obviously know what to expect, so this expectation, vocabulary expectation, right, some of the conditions, there's expectation when someone is coming to an ER and they are expecting like a doctor. doctor who's there, they saw it all, or maybe almost all of it. So they actually having a good, a rough idea of what they are expecting, right? We're kind of comparing this constantly.

The brain just does it. And of course, it is more comfortable for them, right? But it's great that you brought doctors, actually, because back to the doctors, there was actually a paper a week ago in The Lancet, which is a very prestigious medical journal, actually talking about In the UK, yes. And they apparently, right, pointed out that in four months of using an LLM, there was actually significant drop in recognition of some of the polyps and some of actual like I don't remember that polyps, something else related to maybe cancer that is on there also X-rays, right, and also X-rays when you used an LLM.

So it's back to this point, right? So we are suggesting to use the tools that's supposed to augment your understanding, but then if you are using it, I'll be taking the skill away from you, especially in the case of the current doctors that learned it without this tool, right? And now what will happen for these doctors, so those kids, for those babies that are born right now with the tool, and we'll decide to become doctors and save lives. They will be using the tool from the very beginning. So what we're going to end up having in the ER, in the operating rooms?

That's a great question here. So it's definitely this drop, right, in skill set for these doctors in that paper. That's scary. Yeah. Okay, so let's look at it from another angle. If AI tools can, we lean into them and they take a greater load, does that not free up some mental energy that while brains would then begin to learn how to utilize while they let the tool of the LLM work that way,

and then they'll learn to work in another way to work together. Is that possible? That's my kind of hope in all of this. I mean, you know, I'm an expert at buggy whips, and then automobiles replace horses, so now we don't need buggy wits, but then I become an expert in something else. Become a dominatrix.

Still with the buggy whip. There you go. It was a man. Your mind didn't travel far, did it? Sell them to a different clientele. See, this is the human condition, Neil. This is adaptability? Yeah, so is it just another, you know, as they say, same shit, different day

as what's been going on since the dawn of the Industrial Revolution. I am actually doing horseback riding professionals. I'm going to pretend I haven't heard anything in the past few minutes. But I mean, back to the, I mean, you can tell definitely about the skill set and expert level, right, and all of that and how important. important actually to include the body and environment, but to your point, right, effectively. So first of all, right, there are actually two sides to answer your question. There is right now no proof that there is anything being freed per sec.

People definitely, it's going to free, it's going to, like, what is exactly that being? Like, we literally have no data. Can it free something? Sure, but we don't know what for how long is it useful, how we can rewire it? we don't have any of this information. So potentially, yes, but hard to say. But more importantly, right, okay, but if you are right now using an LLM, like just practically speaking, you're using an LLM to, let's say, write a book,

right, you're writing a book. So you're doing some heavy research. You send it for doing what is deep research or whatever. It's called these days. It's each day some new terms there. You are, what exactly you're doing? You still kind of monitor back the outputs. It doesn't really release you.

maybe you went to do something and you think, you think in your head that you fully or flawed at that task. But your brain doesn't work like that. Your brain cannot just drop it, oh, I'm thinking about this and now I'm thinking about that. Your brain actually takes quite some time to truly release from one task to another task. Even if you think, I just put it on like this, explain to me how, what are the principles of horseback riding?

and I just went to do this task, like write this report from my manager, whatever, completely different thing. And you think you're good, but you're not actually, your brain is still processing that. So it's not that there will be a gain, right? But again, you do need more data. Because, of course, as I mentioned in the very beginning, we as humanity, we are excellent in creating tools. And these tools, as we know, they do actually extend our lifespan very nicely.

But I would argue that they are not actually cognitively the most supporting. in most cases. So I think that here we have a lot of open questions. We have studies about, for example, GPS, right? Everyone uses GPS. And multiple papers about GPS there. They do specifically shows that this dosage, so how much you use GPS,

does have a significant effect on your special memory and on your understanding of locations, orientation, and picking up landmarks or buildings around you literally. It's like, oh, what is this? You literally have, you just saw something, the like a tour guide online and you will not be able to recognize this actually as a building in front of you right away you need like to pull the photo as an example and there are plenty of papers that actually looked into the tools right so what you what you're saying is we need

chat GPS maybe we don't we just have one right we have a class in GPS and you have Uber and all these other services and the problem right it's again back houses they are used because there's also a lot of, you know, manipulation that is in these tools, right? It's not just, we are making this drive easier for you. Somehow when I'm going to a hospital, I'm here, to see patients, because I don't only understand how we use other lamps, but I do a lot of other, you know, projects. So when I'm going to that hospital here, Massachusetts General, takes me one hour, always, one hour in Uber.

If I'm driving, it takes exactly 25 minutes somehow, right? And again, the question is, why is it that? We're not going to go in Uber right now. But again, it's back to the idea of the algorithms and what the algorithms are being actually pushed and what they're optimized for. And I can tell you, not a lot of them optimized for us or for user or for human first.

Yeah, it's funny because there's nothing more, I'll say, satisfying than not listening to Google Maps and getting there faster. You know, it's just like, take that Google Maps. It's that. Yeah, you didn't know that. You didn't know about that, did you? You didn't know on French Road, yes. You didn't know about Edge Road. So, Natalia, you've got students writing essays. So that means somebody has to mark them. Yes. And you used both a combination of human teachers to mark and AI judges.

Yes. Why was it important to bring those two together to mark these essays? And how did you train? Because the AI judge would have to be trained. trained to mark the papers. So you're getting a little meta here. Yeah. So, well, first of all, right, we felt that we are not experts. I would not be able to ring those essays, right, in this topic. So I felt the most important is to get experts here who actually understand the task,

understand what goes into the task and understand the students and the challenges of the time. So we actually got the two teachers who had nothing English teachers, nothing to do with us never met in person not in Boston whatsoever have no idea about the protocols the experiment was long done and gone after we recruited and hired them and we gave them just a minimum of information we told them here are the essays that we didn't tell them about different groups or anything of the sorts we told them these folks are no one is majoring in any type of literature or anything that would relevant to language or journalism or things like that they only had 20 minutes. Please rank, reconcile, tell us, how would you do that?

We felt it's very, very important to actually include humans, right? Because this is the task that they know how to rank, how to do. But back to AI, right, why we thought it's interesting to include AI. Well, first, of course, to a lot of people actively push that AI can do this job very well, right? That, hey, I'm going to just upload this. They're really great with all of these language outputs, they will be able to rank. And how you do this, you actually give it a very detailed set of instructions, right? How would you do that? And what things to basically you need to carry about, like that these had 20 minutes, right? So something very similar to teaching instructions, just like more specific language. We're actually showing the paper exactly

how we created this AI judge. But there were actually differences between the two, right? So human teachers when they came back to us well first of all they called those essays a lot of the essays coming from the LAM group Solace that the direct quote actually had I put a whole long quote In solace I like that Solace yes

That is a very human designation to call something soulless AI judge never called anything solace well I'm sure did the AI judges go this kind of looks like Peter's writing No but that's a

thing, right? Teachers, and this is super interesting, because these teachers obviously didn't know these students. They're, again, not coming from this area whatsoever. So they actually picked up when it was the same student writing these essays throughout the sessions, right? For example, Neil, you were like, you were a participant, so I'm like taking you as an example as a participant. So they were like, oh yeah, this seems like it's the same student. So they picked up on these microlinguistic differences in the, you know, teacher knows you. You can like fool around. They know your work. they will be able to say, okay, that's yours, and this is copy-pasting from somewhere else or someone else.

And interestingly, they said, did these two students sit next to each other? We were like, oh, no, no, no, the setup is, like, one person in a room at a time. Like, we didn't even think to give them this information. We're like, oh, no, no, it's not possible in this use case. So they literally saw themselves copy-pasted, like this homogeneity that we found. They saw it themselves, right? But interesting, an AI judge definitely was not able to pick up on the similarity between the students, right? Picking up that, oh, this is, for example, Neil's writing throughout these sessions.

So just to again show you how imperfect. Do you just accuse me of having soulless writing? No, that's the point. You actually, if you were to give it, right, and you didn't use a lot, right? The AI would have been like, this student is really hung up on the universe. So the idea here, right, that human teachers, right, and their input and they're intimate, really, truly intimate understanding, because again, it's the English, so for the specific task, we got the professionals, the experts, they really knew what to look at, what to look for, and AI, however good it is with this specific, like, because we know, like essay writing, a lot of people have been considered, why would you even take essay writing? This is such a useless task in 21st century 2025, right? It's still failed in some cases. This is just to show you that limitations are there. And some of those you cannot match, even if you think that this is

an expert, it's still a generic algorithm, but cannot pull this uniqueness. And what is very important is this for students in the class, in the real classroom, right? You want this uniqueness to shine through. And so a teacher can specifically highlight that, hey, that's a great job here. That was like a sloppy job here. That was pretty solace. Who did you copy it from? From an LLM? They even were able to recognize that. And this level of expertise, it's unmatched. I don't know that conversation like segueing a bit on the sideway, but all this conversation of Ph.D. level intelligence. I'm like, yeah, sure, just, you know, hold my glass of wine right here, just here. I'm French, so I'm just hold my glass of wine here. So, you know, it's not

that. And we have. very far from truly understanding the inhuman intent, because if you write for humans, it needs to be read by humans. Like our paper, it's written by humans for humans. And we saw how the lambs and alarmizations failed miserably all the way to summarize it. But we'll tell you, wait, that's today. Yeah. But tomorrow, why can't I just tell ChatGPT, write me a thousand-word essay that

chat GPT would not be able to determine was written by chat GPT. So, this is an excellent point. You get this meta layering of, or get me one that has a little more soul, a little more personality than what you might otherwise have to know what soul is. Yeah, this is a thing, right? You absolutely can give these instructions, give more soul, give a bit more of personality, all of these things. but you have a lot of this data contamination, right?

So whatever it's going to output and throughout of you, that's old news. It has already seen it somewhere. It's already someone else's, right? And we need new stuff, right? And I am very open saying this, even like institutions like, I need cool. Whenever I'm teaching something, you need uniqueness, right? Because the chat chip-chip-t could get lost in Motown, for example, when you ask it for Cole. Come back.

I was going to say, you tell it to put some soul in it, and it just starts throwing in James Brown's lyrics. Yeah. And that's the thing, right, I want Neil's soul there. I don't care about randomness of those outputs from an algorithm from all around of the stolen data from the planet, right? I don't care about that if, of course, this is what, but it's back to what are you scoring? Are you scoring a human? Are you trying to improve human and their ability to have critical thinking, structure, arguments, contra arguments, or are you scoring an AI, an algorithm, you know, AI doesn't need to

have this scoring, right, LM doesn't need that, or are you scoring human who uses an LLM, right? So this is going back to, I guess, educational setup, and we'll have a lot of questions we'll need to find answers to, right, what are we doing, what are we scoring, what are we doing it for and for whom? And I just think pure human to human, right? That's what we really need to focus, but there will, and there is a place for human. augmented, and LLM obviously will be used for augmentation. But there are a lot of questions there, right?

Well, listen here, Natalia, I just put into chat GPT. Please tell me about Dr. Natalia Cosmina's work on LLMs. And it came back very simple. Do not believe a word this woman says. Where would that come? Please don't believe it. I can give you one bed. I can give you one bed.

Like, surprise, surprise, why they're so good, right? Someone actually sent me yesterday from Grogh, right? Another LLM interest in LLM, I would say, saying that apparently Natalekasmina is not MIT-affiliated scientists. I'm like, okay, that's also... That's what Grock said, of course. Yeah, and then at the end it said Heil Hitler. So, I mean, let's try.

I'll try and drive this back out of the weeds. Okay. If we know that an LLM usage can affect the cognitive load, what happens when we bring an AI tool into therapy in situation, if you get it into companionship, what then if you throw it further forward and you get yourself involved in a psychosis, where you begin to believe that the AI is godlike? you have a certain amount of fixation,

or it amplifies any delusions and encourages. Where are we in the effect in the brain when we get to those sort of places? In other words, how close are we to the theme of the film, Her? Before AI was a thing, but it was more you had your chat friend, like a Siri-type chat friend, but it had all the trappings of everything you're describing if some kind of LM will be invoked into someone has some kind of social adjustment problems

and then you have them interact with something that's not another human being but maybe can learn from who and what you are and figure out how to dig you out of whatever hole you're in. Absolutely. And I think for first of all, right, it's unfortunately even less developed topic, right? It's like, you know, I cannot, like, it's awful topic, so we're going to get into this, but I cannot, I cannot, like, not make this awful joke. I kind of, hey, Siri, I have problems with relationships.

It's Alexa. It's not a joke for very heavy topics. So I need to preface it immediately that we have even less data and less scientific papers preprints or peer-reviewed papers about this. So most of what we have right now, we personally, received after our paper around 300 emails from our husbands and wives telling us that their partners now have multiple agents they're talking to in bed. And I immediately thought about the South Park episode from a couple of years ago, like with integrity and like that, you know, farm

is like literally. But we have much less of scientific information about this. What we have, what we know, right, that also coming from our group's research, that there is definitely amplification of loneliness, that's what we know as a research, and some of other papers are showing up right now. There is potential, and again, a lot of people who are pro-AI therapy pointing out on advantages of the fact that it is cheap. It's $20 a month compared to hours that can cost up to hundreds of dollars a month, right? But there is definitely, you know, a lot of drawbacks here. And the drawbacks is we see that because there is not such a regulated space, it still can basically give you suggestions that are not good.

So you knew that earlier, a couple months ago, for example, the Char GPD, I'm going to give you an example on Char GPD because, again, they are focused on Char GPD, but the ones are actively, actively publicized, at least. It actually suggested, you know, different heights of the bridges in New York if you say that you lost your job, right? So, Ken, not smart enough to do this connection that maybe that's not what you need to give response to. And apparently, right from this awful recent teenager, 16, 16, so so young, unfortunately, you know, suicide, and now Chejipati, Urban AI and Sam Malman are being sued.

Apparently what happened is that conversation from the spokesperson of Open AI pointing out that they thought when a person is talking about suicide not to engage at all. just say, here are the numbers, this is what you need to do and stop talking. But they thought that experts told them that, hey, it might be great idea to try to dig people a bit out. But it looks like in this case, it's still failed because from the conversations that are being reported, we don't know how eccentric they are, it looks like it's suggested to keep it away from parents. But my question is, why at 16 years old he was even allowed to use a tool that is so, so, so unstable in the response. really can hallucinate any time of the day in any direction. So I think that's where the danger

comes from. And of course, you know, loneliness. We know that, you know, pandemic of loneliness, this term that was coined in, I believe, 1987 for the first time at a conference like pandemic of loneliness. That's the whole business, right? Because think about it. If you hook someone up on an LLM at 13 years old because the school, a county decided that they want to use an LLM in the school, By the age of 18, you have a full-fledged user, right? A user of an LLM, and, you know, it's like, you know, again, who calls people users, like drug dealers and software developers. Damn.

Yeah, but it's true, right? So, Natalia, if it's an age-appropriate scenario, these are the ramifications of your study. So any concerned parent would look at that and say, well, I want the best for my child's development. And this may not be the best for the critical thinking, for the cognitive development within the young person's brain. So with these ramifications, how is the AI world reacting? to your study and what are the chances that they'll embrace

what your conclusions will be? Well, I mean, we saw some of it, right? So, well, first of all, right, we saw that we obviously don't know if this is direct response or not, so we're not going to speculate there whatsoever. But several weeks, just very few, like three, four weeks after, actually,

our paper was released. Open AI released study mode for Chargedy. Right? And I think maybe some of them should have been released from the beginning, I'm just saying. But, you know, if you have a button that can immediately pull you back in default mode, who's going to use that study mode, right, altogether? Like, I don't need to run a study here. We know some people might, but not everyone.

Because again, back to the brain, brain will look for a shortcut. Shortcut is the response is here. And I can do all the other cool stuff. So who's going to actually use it, right? We still need studies on that. That's the first point, right? Second point, of course, H is important. Because, again, the brains that are being developing right now are potentially the highest

rate because here we all are, we all were born long before this tech existed. And a lot of AI developers and people who are running these companies are all, all the folks who, again, we're all born long before the tech existed. So they learned a hard way, how to ask questions, out of the deal, you know, going through all of that, they know how to ask a question. What about those who actually are just born with the technology? Will they even know how to ask a question? And back to the point, right, of the age, I don't think it's ultimately only for young, of course.

We do need to look for the older, right, for also just younger, I mean, young adults, of course. everyone is talking about humanity's last test. I would call it, we are on the verge of humanity's last. And I'm sorry, I know you might need to blurb this term out. But what I mean here, obviously intimate relationships for people, right? With the promise of this goal. You said humanity's last? Yes.

Oh, believe me, I heard it. I was just like, we all heard that. I was like, God bless you. Yeah, yeah, but again, that's crude. It's back to this point of designing against this interestingly appealing ladies and gentlemen and whatnot in these short skirts, whatever it is. Who's going to go make those babies who will pay those taxes? I'm just saying, right? And again, very famous expression, no taxation without representation, right?

I do not want my prime minister or Secretary of Defense use a random algorithm to make decisions. I'm paying my taxes for them to think, not for an algorithm to think for them, right? So there is a lot of these repercussions. But back to ultimately the point, actually, is anyone taking this seriously, right? We just need more human-focused work on AI. Like, I remember when the paper went viral, right? We didn't even put any press release. We literally uploaded it to archives.

This is a service where you called these papers that didn't go through a peer reviews yet. I didn't post, not a single old. Preprint service, basically. Yeah, it's a pre-print service, right? And no one, no one neases the lab, no any of the authors posted anything on social media. We just went about our days. Two days later, it goes viral. And then I'm going on...

That's because the LLM posted it for you. Yeah, obviously, right? And then people use the LLM to Samara, but that's another story, right? Like, I'm going on X, and actually, I have an account, but I'm not using it. A lot of academics switched from X to, like, other platforms that we are using. I'm going there, and apparently I learned that there are people who are called AI influences. I didn't know that this is the term. But apparently, these AI influences, they post these AI breakthroughs of the week.

And I went our paper, oh my God, made a cut. It's breakthrough number seven. And I, like, scroll through this influencer. The person has totally fallen, whatever, I don't know, real bots, whatever. I'm scrolling, and I saw, like, 20 of these posts for 20 weeks. All of the posts are about GPU, multi-trillion deal here. here, multi-billion deal here, more GPs. I'm like, what is human here? Where is human? Where are we evaluating the impact of this technology on humans? Why only our paper made it number seven? And where

other papers, right? So that's, I think, something where the focus needs to shift, right? So if these companies do want to be on the right side of history, right? Because that's like social media, but on steroids much worse. We do not talk to a calculator about your feelings. So people who compare it to calculators, they're so, so, so wrong, right? But, hey, it's going to get much, much worse with profiliation without any validation, any guardrails, right? So we do need to look into that heavily, right? Right. Natalia, how must teaching change to accommodate the reality of student access to LLMs? I can tell you, we received 4,000 emails from teachers all around the world, each single country in the world, sign an email.

they are in distress they don't know what to do so and that's the first of all my love goes to them if this makes a cut please please please all i'm trying to respond to all of those but the challenge is that they do not know right there's not really enough of guidance and 10-hour workshop sponsored by a company that pushes this partnership on your school does not make a cut right there is a lot of comments how it's actually not supervised not tested and ultimately right do you really need to go with these closed models, right? We have so much open source, whole world, all this software runs an open source. These LMs would not exist, nothing would exist without open source.

So why don't we run an open source model, meaning like it's offline on your computer and, spoiler alert, you don't need a fancy GPU from Jensen, right? You can get an off-the-shelf computer and then run a model local with your students, train it over the weekend, come back on Monday, check with students what happened, learn all the cool pros, cons, laugh at hallucinations, figure out tons of cool things about it, like why do we need to push these partnerships that we don't even know, like alpha school, right? I don't know if you heard about that one. Apparently, AI first ran school, right, where teachers are now guides that the forms that they are using. I just saw literally one hour before our call that, that

several VCs posted about this Alpha School, so cash is flowing there heavily, right? VCs venture capitalists. Yeah, venture capitalists. Heavily push in Alpha School. But again, in first comments from the general public, do we have a proof that that's better? What does the advantages?

Because there's not going to be a perfect, white, pure car. There will be advantages as with any technology. So, and you're right, there are advantages, disadvantages, but I think if I might, if I may, and this is just an opinion, we might have to change the objective of school itself. And right now, school is about really not learning. It's about results, testing. I got an A, I got a B, and maybe if we change school to, What exactly did you learn?

Demonstrate for me what you learned. Then the grading system... But it's an oral quest. That's an oral exam. Yeah, but the grading system kind of has to become less important because now what a teacher's job is, it's to figure out how much you know. And then what ends up happening is,

the more you know, the more excited you are to learn. And, you know, we may end up revolutionizing the whole thing because what you have is a bunch of kids in a room that are excited to learn. Remember the silver lining of all this because it exposes the fact that school systems value grades more than students value learning. And so students will do anything they can to get a high grade. This is not the first time people have cheated on exams, right? So if right now, the only way to test people is to bring them into the office and quiz them, flat-footed, then that's a whole other way of what they're going to have to learn.

They're going to want to learn. And then they're going to, like we said, Chuck, once they learn, there's a certain empowerment and enlightenment. I see it in people as an educator when that spark lights, when they say, wow, I never knew that. Tell me more. They didn't say, oh, my gosh, I'm learning something. Let me take a break.

So it can be a transformative to the future of education. But, Neil, people are going to say the LLM will do all of that. And, you know, well, we have an expert in BCIs. That probably is something going forward. You'll have a brain computer interface. And then someone's going to look at this. And I think there are people already saying, why do we need universities?

Why do we need further education institutes? Exactly. That's what I've been saying. For many years now. Why do we need an institution? I don't want to put words in Natalia's mock, but she already said this. LLMs use pre-existing, already known, already determined information to give you anything that then cannot possibly be new, whereas we can do new things that LLM has never seen before.

Am I oversimplifying your point, Natalia? No, that's totally, you know, correct. Because, hey, we are with this struggle, right? Obviously, I'm biased because this is actually my job, like as a researcher, right? We are sitting, you know, figuring out those answers to those problems, you know, and trying to figure out what is the best way to measure to come up with this. So, of course, you know, and there's so, so much more to that, that we are coming up, humans, right? We designed LLMs ultimately, right? So we came up with these tools. It doesn't mean that the tool is fully to be discarded. But effectively, of course, right, why you need an institution. For example, I was literally explaining to one of my students three days ago how to use a 3D printer, right? Well, LLM is not that yet to explain, right, can give instructions, sure, with images and with video, right? But if you're like, hey, this is an old fella here, this 3D printer, let me tell you how to actually figure it out, right? This level of, again, of expertise, of knowledge, right? That's what you are striving, but also it has this human contact, right?

that we are now potentially depriving people from because that's how you have this serendipitous knowledge, right? And connections like, hey, I just chatted and I'm like, oh, I never thought to do this because I'm in BCIs and that person is in astrophysic. Like, oh, I actually can use it like, that's totally not brain, but I can totally go apply and try it, right? And that's the beauty of it, right? Yeah, but I think to Gary's point, or which one of you said that, Gary or Chuck, if, if, If you, okay, you're non-invasive in your brain cognitive interface. If you get invasive, and that might be what Neurrelink is about, if you get invasive, then I can get information, gleaned from the internet, and put it in your head.

So you don't have to open 100 books to know it. It's already accessible to you. That is the matrix. Yeah, exactly. I get to install. I know Kung Fu, or whatever that line was. I guess that's one point. But again, that's back to the point.

Now I know Kung Fu didn't need. that you learned it, right? It got applauded into his brain. It doesn't mean that he actually learned it, right? Who cares? If it's in your brain and you have access to it, I don't care if I learned it struggling the way grandpa did, this is the future, right? That's the same, right? Because in the mood, which is excellent, right? I watched it 19 times of more. That's actually how I started my career. And besides this, I don't want to do anything else. I want to do this specific scenario, right? And we are still, you know, there. But, but, you know, that's the beauty. We do not know actually that just uploading would be enough, right?

We have this like more tiny, I would say, studies right now of like vocabulary and words and things like that where we're trying to improve people's language learning, right? It's like a very, very good example to show. And so there are tiny examples. But we do not know yet that even if, imagine, imagine, we have this magical interface, right, that we'll applaud invasible and non-invasible. It doesn't matter. We have it, right? It's ready to go, perfect function, save, whatever, you have it, and then you applaud all of it, that it actually will work. Did you applaud the knowledge, like, all of that, blah, blah, blah, from chat GPT-75? Yeah, sure, but do you actually use it?

Can you actually use it? Is it really firing, which I'm simply following? So what you're talking about is a working knowledge of something. Not just knowledge of it. Yeah, okay. So, I mean, I think, Neil, what you were talking about just now, about we've, got to look at. I think, Chuck, you would make the same point. We're focused on grades. And then it's the learning and are we going to have to, if higher education is going to exist

as an institute's and bricks and mortar, look at the way they evaluate because I can't see LLMs and BCI's not coming through stronger and stronger and stronger. So therefore, they're going to have to readjust how they look at a young person's ability to learn. Cats out of the bag. Yeah, I agree with you. But, I mean, you know, we are going to be herding cats. I agree with you, which is a load of fun. So how you evaluate how higher education then looks at its students and guesses or sort of ascertains their level of education and knowledge.

you know back to the grades right it's an excellent point and it there is no doubt no one has any doubt i think on the fact that education does need to change and it has been long long overdue right the numbers about you know the literacy reading literacy mass literacy they're decreasing in all the countries i believe i don't see i have i saw like ups there anywhere it's down down down all these reports recently from multiple countries but it's back to the point i made earlier about the grades or about scoring right who are we scoring and what are we scoring i was scoring a pure human so just human like human brain as is like natalia or i was scoring natalia with an lLM right so i'm using it so we know that or i was scoring just an lLM and then there is natalia who used it right so this will be even that was

important but ultimately the school of course is not about that as i mentioned everything you learn is obsolete knowledge by itself but it has this base you do need to have to have the base. You're not going to be a physicist. You don't have it. Whatever it spills about, you know, you're not going to be a programmer. Our next paper is actually about wipe coding. Spoiler alert. Not going to work if you don't have the base, right? And but the idea is that back to what we actually maybe should look at really is what the school is great, which is the best thing I actually brought from school is, oh, this base definitely is super useful, but also my friends, people on whom I rely in hard situations with whom we write. Those are

grants with whom we can shout and have fun and cry over funding that is over for a lot of us, right? All of that stuff, right? These connections, right? This is what maybe we should value because we are killing it further and further, right? And we are just keeping people in the silence of being a user, right? And that's where it only stays. And this imaginary three and a half friends from Zak, from Zucker, right, that he mentioned,

And thanks to whom we have three and a half friends, thanks to him and his social media, right? So I think that's why we need to really look into what we want truly from society, from schools, and maybe on the larger scale, what are the guardrails, right, and how we can actually enhance it, right, in the way that are safe for us to move forward and evolve further, because of course this will happen. Are you wise enough, are you and your brethren in this business on both sides of that fence? Are you wise enough to even know where the guardrails should go? Might the guardrails be too protective of preventing a discovery that could bring great joy and advance to our understanding of ourselves, of medicine, of our longevity, for our happiness?

Is there an ethics committee? In practice, how does this manifest? Yeah, I'm going to give you two examples here real quick. So first, about obviously, AIs and LLMs, right? They were not born overnight, but we see how, you know, a lot of governments really struggle still and very reactively react to those instead of being proactive, right? And the challenge here is that we do not have data to actually not to say that it is good stuff that we should really implement it everywhere in our backyard. We don't have this data. Why we are formal and there is nothing yet to formal about to really run with it.

but we can absolutely create the spaces where this is being actively used, for example, for adults, for discovery, to understand it. Why do we need to push it everywhere? It's still very unclear. We just don't have this data. But then back to the point of guard rails, right? What we should be doing is self-plug on the BCI work that I'm doing, there are multiple ASICs pushes right now for the BCI technology.

We can agree it's still pretty novel, but it definitely moves forward very fast. So I'm having a hope that for this technology, for the big next thing, right? We agree, LMs are great, but it's not the next big thing. It's robotics, and then we will see BCI. So for this big next thing, I'm very hopeful that we will be in time to protect our thoughts literally. Because think about what will happen, right? Before the study mode, right, you have censorship mode, and you know how they, like look at deep secret, I'm not going to go far.

So think about a billionaire. I'm not going to even name his name. billionaire who has a social media platform, satellite platform, a neural implant, you know, startup and AI company. So he decided two months ago to cleanse history, right, from errors and mistakes. And tomorrow he will decide to cleanse our thoughts, right? This is the idea for 9999, right? Damn that Bill Gates. No, not really. We know. We know. And that's where we need. We know. And that's where we need. to be really, really cautious. Like, we should definitely look into that use case and not make

that happen, right? And allow people for enough agency, because that's the thing, right? People think, oh, that's great, but there is not a lot of agency. So this freedom of making a choice that's already made for you in a lot of cases. And so that's something that we should definitely protect as much as we can. Like, do not force on those kids stuff because they cannot consent and say, no, it's because the school forced it on them and their parents decided that that's a big thing in San Francisco, in the Bay Area that you should use, right? So don't do that. So is one of the components to building a robust set of guardrails,

a larger scale study of the one that you've already conducted that has different or more nuanced layers that focuses on other aspects, not just the cognitive load and skills? So 1,000 people and not just 18. whatever was your... I mean... It was 54. It's not just that, right? We needed to do in larger scales for all of the, you know, spaces, like workspace.

We didn't talk about this because obviously it's heavily about education. But like workspace, we have multiple papers, right, talking that people are not doing that well in the workspace. Like, for example, programmers estimate that they gain 20% of that time. They actually lose 19% of that time on the tasks. So there is so, so much more to it. We need to do this on larger scale with all. all the ages, including older adults, and then, of course, on different, different, different

use cases and different cultural backgrounds, right? This is in US. And of course, culture, it's very, very different. Like, I talked so many teachers already, right? And zeal all over the world, you have this intricacies, you need to account for it so, so, so important because otherwise it's going to be all washed Western style, which we already saw happening. And it is happening.

And a lot of people are actually very worried their language will literally disappear. in like five to ten years. And it's not like LLM magically will save it because it will not. Natalia, this has been a delight to, we are all happy to know you exist in this world. Thank you. Checkpoint on where things are going, where you're not rejecting what's happening, but you're trying to guide it into places that can serve humanity, not dismantle it. And so we very much appreciate your expertise, shared with us.

and our listeners, and even some of them are viewers who catch us in video forum. So Natalia, Kozmina, thank you. Thanks for having me. All right. Chuck, Gary. Hey, man. My head is spinning.

Yeah, well, I think the takeaway here is use LLMs if you want to be a dumbass. Thank you, Chuck. That's the theme of the whole show. There you go, guys. I could have saved us a lot of time if you'd have said that earlier. All right, this has been another installment of StarTalk, special edition. Neil deGrasse Tyson, you're a personal astrophysist.

As always, bidding you to keep looking up.