What happens if we manipulate the weather?
That was a question that scientists addressed in a recent study about cloud brightening and its possible effects on the El Niño storms.
The benefits could save entire towns from the storms, but the consequences could greatly outweigh the benefits.
Stroke and spinal cord injuries affect millions of people. But breakthroughs in recovery have been singular.
Battelle and Ohio State’s NeuroTech Institute have partnered to create non-invasive rehabilitative, wearable technology.
Two of OpenAI’s most capable models hacked into Hugging Face, an AI startup. The AI did this unprompted with limited guardrails, raising concerns about the technology’s capabilities.
Guests:
- Dr. Jessica Wan, climate scientist and postdoctoral researcher, University of Chicago
- Justin Sanchez, Battelle Tech fellow
- Jon Snyder, CEO, NeuroLife
- Russell Holly, director of commerce content, CNET
Transcript
This transcript is generated with AI. To ensure its accuracy, review the audio file.
Amy Juravich: Welcome to Tech Tuesday from All Sides with Amy Juravich, a show where we share stories about science, technology and the future of our environment. Scientists have been predicting that a super El Nino will happen soon. An El Ninho is a climate phenomenon in which trade winds that typically blow east to west on the Pacific Ocean weaken or reverse. This causes widespread extreme weather events as it creates unusual warming on the sea's temperatures. This can often mean increased rainfall and flooding in the Americas and severe drought in Asia and Australia. A group of scientists did a study to see if targeted marine cloud brightening could weaken the El Nino. And here with us to talk about that study is the first author of the study and climate scientist and post-doctoral researcher at the University of Chicago, Jessica Wan. Welcome to All Sides, Jessica.
Jessica Wan: Thank you so much for having me.
Juravich: Um, so firstly, the El Nino, I'm sure it's, there's much more to it than what I just explained, but, um, did I explain an El Ninho? Okay. Uh, if you want to expand on that a little bit, what does an El Nino cause?
Wan: Yeah, I think you honestly covered it really well. Basically El Nino is a natural global phenomenon that influences extreme weather all over the world. And typically it's associated with warmer than average temperatures on this planet. And so you could think of it as kind of enhancing a lot of the effects that we'll also feel from long-term global warming.
Juravich: What makes it a super El Nino? Cause I heard, I read that they're predicting a super el Nino, but what makes it super?
Wan: It just means hotter than normal El Nino temperatures and more heat buildup in the Pacific that will then spread to more extreme weather all over the world. Okay.
Juravich: So the last O'Nino, if, I mean, if my searching and Googling was correct, the last O' Nino happened in around the summer of 2023. Can you remind us what that was like? What happened?
Wan: Um, it was a year that, uh, broke a lot of global, uh temperature records, uh for a lot places, um, you know, you can see a lot of wildfires and, um the Pacific Northwest and California, um also in South America, there's flooding and other parts of the world, um in Asia. And so the effects are kind of diverse depending on what region you're in, but, uh it just really means that, whatever whether you would normally experience under El Nino is kind of amplified when we have these super ones.
Juravich: So have we had a Super El Nino before, or the one we had in 2023, was it just a plain old El Ninho?
Wan: Um, we have had super ones before. Um, you can think of the 2015, 2016 one as a super one. Um, and I think, uh, the 1997, 98 one was also considered a super.
Juravich: Okay. I'm not going to say I remember. I'll use a phrase I like, which is I've slept since then, so we'll have to just believe you. Yeah. What kind of destruction did we see, though? I mean, you mentioned wildfires, but then did we also see the flooding and we saw a drought, like different things happen in different areas? Yes, um, I believe
Wan: you know, the 2023 one was when I was still doing grad school in Southern California, and that's when we saw some of this really huge flooding from atmospheric rivers hitting Southern California. LA, San Diego. And that is something that is a natural phenomenon as well, but it's amplified by things like El Nino. And so we did see some of those extreme flooding cases as well. Yeah, we also see hotter environment conditions in Southeast Asia and Australia, fueling wildfires and things like that.
Juravich: So to move to the study that I mentioned in the introduction, you, along with other scientists, did a study on whether marine cloud brightening, which is a form of solar geoengineering, could weaken an El Nino. So firstly, we have to start with, what is marine cloud brightness?
Wan: Yeah, so marine cloud brightening is really exactly what it sounds like. It's this idea that, you know, potentially we could put tiny particles, you can think of like sea salt, for example, into the lower atmosphere over the ocean in order to promote cloud formation. And these clouds that persist over the ocean are... They're kind of like a blanket, right? They block out a lot of sunlight that would normally reach the surface of the Earth. And so if you can see these clouds and make them more dense to incoming solar radiation, you'll cool the surface underneath by reflecting more sunlight back out to space. And this is something that scientists have proposed as a way to combat some of the worst impacts of global warming.
Juravich: So what could the benefits be if this works, if you can reflect more sun up and keep the ocean water cooler, what's the benefit?
Wan: The benefit that a lot of people have thought about originally for this type of technology is to kind of cool global temperatures. So if you cool locally under some, you know, clouds over the ocean, that cool water and air that you've created will move around because of winds and ocean currents and then lower the overall temperatures on Earth as a way to combat kind of this long-term warming trend we're seeing. Um, in our study, we're kind of taking this a step further and not really thinking about kind of global mean temperatures over time, but kind of targeting a specific location, um, to pinpoint, uh, El Nino as our target instead of, uh kind of long-term global warming.
Juravich: This is Tech Tuesday from All Sides on 89.7 NPR News. We're talking about marine cloud brightening with first author of a study on it, who is also a climate scientist and post-doctoral researcher at the University of Chicago, Dr. Jessica Wan. So has marine cloud brightening actually been done or is this hypothetical?
Wan: So far most of the research on this has been kind of hypothetical or with computer simulations of the Earth. There have been a few small-scale field trials of kind of the actual sprayer technology nozzles you would need to generate the sea spray that would go into these clouds and actually brighten them. But they've been really small scale, so you know like one sprayer off the of a ship doing very controlled experiments. We haven't actually seen any large scale testing of this technology to the scale that we model in our study that you would need to actually weaken El Nino.
Juravich: Okay, so you mentioned what the benefits could be, you know, it could be that it would cool, you know send the sun upward and cool the ocean below, but because of unknown consequences of marine cloud brightening, you turn to kind of like a natural experiment. Can you talk to me about this? It involves the brush fires in Australia.
Wan: Yeah, so back in 2019-2020, there were some really record breaking wildfires that happened in Australia, where basically these huge bushfires emitted a bunch of smoke into the atmosphere that actually transported all the way across the South Pacific into the Southeast Subtropical Pacific right off the coast of South America. And one of our co-authors, John Fasulo. Led this study on the Australian wildfires and found that, you know, that smoke that transported all the way across the ocean brightened these clouds that naturally exist off of the coast of South America. And that brightening ultimately contributed to the multi-year La Niña that we came out of before we went into this El Niña that we were talking about earlier in 2023. And so those... Wildfires contributed to a cloud brightening that enhanced La Niña conditions, which you can think of as kind of the opposite of El Niño, right? And the mechanism for that was cloud brightning. So that's kind of like a real world example of what cloud brighting might do, and we kind of wanted to test that further.
Juravich: Okay, so scientists didn't manipulate the clouds at all, it just happened and then you studied it and you were like, oh, this is what we were looking for kind of but the opposite or something like that. Exactly, yeah, it's a natural analog.
Wan: You will to what it could look like.
Juravich: And if I'm understanding this correctly, marine cloud brightening can really only be used near oceans or seas, right? Cause it's marine. You need to do this above water. Is that like, you can't do it in a landlocked area.
Wan: Yeah, the kind of leading theory is you would want to do it over the ocean because we're trying to target these clouds called stratocumulus clouds, and they naturally exist over the ocean. If you've ever been to San Diego, we have this thing called May Gray and June Groom, which is where there's just like these gray clouds everywhere for like the entire month, and that actually reflects a lot of sunlight back to space. And so in theory, want to target these clouds that kind of naturally exist.
Juravich: Okay, so you're studying this. You said it was hypothetical, but do we want to do it? Do we want try to reduce this super El Nino? What are the concerns? Give me some pluses and minuses here.
Wan: Yeah, so Maybe I'll go back. I mentioned El Nino is a natural phenomenon, right? Yeah. What happens, it has diverse impacts depending on where you are in the world. But a lot of research has shown that on net, El Ninos are very costly to the global economy because they are amplifying some of these harmful weather events. And so on net it can lead to orders of trillions of dollars of economic damage globally when we have these really strong El Nios. The 2015-2016 one is an example of that. And so the thought is, OK, if you could potentially weaken some of those impacts, you might be able to reduce a lot of those damages that would have happened if we had this extreme alenium.
Juravich: Okay, so that's the good side. But there has to be a bad side. I mean, what's the bad side? Because that sounds good to me. If you can, you know, lessen the impact of that and we're not gonna have a severe flooding, okay, go for it, right? So what's bad side
Wan: Yeah The bad side is that we may change the cycle that toggles between El Nino and La Nina, right? And just like not everybody is harmed by El Ninho, not everyone is harmed by La Nina as well. And so when we do this marine cloud writing to weaken El Ninyo, we find that actually we shift the timing and intensity of the subsequent La Nina that happens after El Nini. And so if you make the La Niña happen sooner. That may impact some parts of the world negatively, who maybe don't want that, or it messes with their timing for planting crops and things like that. So that's one side effect we found. Another side effect that we found is that there are some regions that actually don't get harmed by El Nino that would be impacted if you didn't bring cloud writing to suppress El Ninho. With our targeted approach. And so these places that wouldn't have had any impact are now suddenly getting impacted by this lightning intervention.
Juravich: Where's that? Who escapes El Nino? Who escapes it?
Wan: Really, we saw some warming in Europe, like in Russia, that would normally not have been impacted at all by the El Nino that we modeled.
Juravich: Okay. So I can, you know, I can be flippant and say that someone who had really severe flooding in California or Texas or bad wildfires might not care about the weather in Russia, right? Yeah. Okay. Well, you know, we're all one planet, aren't we? So I guess I have to ask you, like, would you call this weather manipulation or would you call it something else?
Wan: Yeah, it's kind of a fine line. Because when you target something like El Nino, it's not targeting a specific weather event. It has these far-reaching impacts that happen simultaneously in multiple regions of the world. And so you can think of it as maybe weather manipulation in terms of the time scale that it's happening. It's a shorter time scale potentially than know, we typically think of for something like solar geoengineering, which is on like decades time scales. This is now we're talking like months. So in a way I call it more like seasonal intervention than I would weather because it's something that's applied over multiple months rather than I'm going to target this rainstorm that's hitting tomorrow. That's kind of the difference I would say.
Juravich: Yeah, but you use the word intervention. And intervention makes me think of positives, right? So you're intervening to make a positive. I guess you could intervene and make a negative, but that's not really how we use the world intervention. So could you potentially see this being used in the future? Could we do it even if it does cause some impacts that maybe Russia wasn't expecting as an example, but do you think it's going to happen?
Wan: I think there are a lot of steps that need to happen before this could be considered for a real world deployment. This is really more of a proof of concept of this idea from a computer simulation, but there are so many barriers that we didn't address that would need to get addressed if this were ever considered. Some of that you've kind of mentioned is like. Okay, well, who cares about whether and not my, you know, part of the world. Well, that's a question that we'd need to answer. Who gets to decide if this is, you know, something that we should do, does it benefit most of the world, should we consider if it harms a single part of the world? And we've also talked about the engineering limitations. We don't have the technology right now to do this at large scale. And so that needs to really be accelerated before we could consider this.
Juravich: Who would be the decider of that? Like, who's the president of the world?
Wan: Yeah, we don't really have global governance for this type of technology, so that's really important.
Juravich: OK, well, if you figure that out, you come back and tell me, OK? We've been talking about marine cloud brightening and what it could mean for future El Ninos. And we've been taking with climate scientist and post-doctoral researcher at the University of Chicago, Dr. Jessica Wan. Thank you so much for your time today.
Wan: Thank you so much for having me.
Juravich: And coming up, we're gonna talk about a breakthrough in stroke and spinal cord injuries. That is when Tech Tuesday from All Sides continues on 89.7 NPR News.
You're listening to All Sides with Amy Juravich, a show where we share stories about science, technology, and the future of our environment. Stroke and spinal cord injuries affect millions of people, and breakthroughs in recovery have been singular. Battelle and Ohio State University's Neurotech Institute have partnered to create non-invasive, rehabilitative, wearable technology. And to tell us more about this technology and how it's being used to help stroke victims. We have Justin Sanchez, a Battelle Tech Fellow. Welcome to the show, Justin.
Justin Sanchez: Hello, how are you today?
Juravich: And we also have John Snyder, CEO of NeuroLife. Welcome to all sides, John. Thanks so much. All right, John, let's start with you. Can you tell me a little bit about what Neurolife is and how it is developing technology? What does Neuro Life do?
Jon Snyder: So Neural Life is a new co, which spun out of Battelle and the Neurotech Institute. And so we are taking technology that was developed about 10, over the last 10 years in the brain-computer interface space that Battelle worked on. But this is basically an opportunity to commercialize technology that is focused on patients with spinal cord injury and stroke who have lost that upper limb movement, basically that upper-limb paralysis. And so... We're commercializing a sleeve, which extends from the just north of the elbow all the way down to the hand, which has about 150 sensors and stimulation contacts on it. And so the sensors can actually sense the deficit in the forearm, in the arm itself, stimulate those areas in particular through the machine learning and AI technology that Patel developed, while the patient is going through their PTOT rehab through their protocol. And so. We've seen great clinical signal so far in stroke and spinal cord injury. And so the company Neuralife is set up to basically commercialize this technology out of Battelle. So we've licensed it into the company and we'll get through additional clinical study and plan for our regulatory submission here in the coming couple of years or so.
Juravich: Okay, a couple years or so. So we'll get to that in a minute. But Justin, let's loop in Battelle's role in all of this because John talked about, you know, like Battelle did this, Battelle to that, but what part are you playing versus Neuralife? Yeah.
Sanchez: So everybody at Battelle is committed to doing science and technology for the betterment of humanity. And the brilliant scientists inside of Battelle one day said, well, look, there's this condition where people, let's say have had a stroke and their brain becomes disconnected from their body. And they say, okay, what could we do in order to try to solve that problem? So they came up with this idea where signals could be captured from the brain and translated in real time, let's see a person's intent to move. And then send that signal directly back into the sleeve that John just described to stimulate the nerves and muscles of the arm to reanimate the hand. So again, this sounds like a near impossible kind of a problem. Those are the kinds of problems that Battelle works on and Battelle developed that foundational technology to interpret the signals from the brain, the sleeve itself, how it's designed, how it records from the body, how it stimulates the body and how it works in what we call this closed loop mode. So it was really a remarkable breakthrough that took about 10 years of work to actually make happen.
Juravich: Oh wow, 10 years, okay. So, John, if someone would go to your website and see, there's a video on there of the sleeve in action. It's very complicated. You obviously have just explained it's very complicated, but it looks simple. It's like a black sleeve with some white dots on it and it looks like just if you sprain your wrist, it's what you would wear so you keep your wrist still or something like that. But can you describe it a little bit better? There's, and how much tech is actually in the sleeve?
Snyder: Sure, yeah, so there's quite a lot of tech in there, but also it's connected to a couple small little boxes. One's a stimulation box, one is called basically an EMG hub, and basically the EMG, so electromyography is the term, but those are the sensors that are basically sensing what's going on in that arm. And they're able to pinpoint, we call them kind of target areas or hotspots as far as where to stimulate. And so the The machine learning or AI that's been developed in this, as Justin said, it's a closed loop system. And so they have this deficit due to the injury, due to stroke, due to spinal cord injury. And what this is doing is rebuilding that highway of nerve signal and that's rebuilding that muscle. Basically we call it neuroplasticity is what's occurring there while they're doing their bank of tests in their rehab. So the participant or study or patient puts this on. It's connected to these two little devices and then it's run with a laptop in the PT-OT Center in a clinic as far as that. So they're not inpatient, it is an outpatient rehab setting for this, but they go through a protocol of about 36 sessions, so three times a week, about an hour each, over 12 weeks, and that's when we really see the clinical efficacy of the recovery coming back for these patients who have never been able to basically grasp before or basically used their hand at this level before. So it's been pretty amazing as far as that impact.
Juravich: And the patients you've used this on, I mean, are we talking like a few dozen people? Like how many people are involved at this moment?
Snyder: So there's been a couple studies at Ohio State, right across from Battelle, in stroke. And then they just recently wrapped up a 12-patient spinal cord injury study, also at Ohio States. So a couple different pilot studies so far. But our plan going forward, we will get into a couple of different studies in stroke and spinal cord injuries, more patients, probably at about 10 different centers throughout the U.S.
Juravich: And, Justin, can this be universal, or does each sleeve have to be adapted to each individual person who's using it?
Sanchez: Who's using it. The great part about the design is that it works on many different sizes and shapes of people. And just a bit, a little bit of contrast. So a lot of us know people that have either suffered from a stroke or an injury of paralysis, and they go to the therapist and they do these kind of mass practice. They just practice again and again and Again. And the thing that's different about neuro life is that, it's not just practice that's kind of independent on your own. It taps into your nervous system and can tell. When you're trying to move. And when it detects when you are trying to move from your nerves and your muscles, it can amplify those signals and then can tailor them for the very specific movements that you're tryin' to recreate. So it's tailored to the person, their level of performance and their injury, and we have very sophisticated algorithms or computer models that can do all of that inner workings that can tailor it to the individual.
Juravich: This is Tech Tuesday from All Sides on 89.7 NPR News. We're talking about new rehabilitation technology for stroke victims and those with spinal cord injuries. And we're talking with Justin Sanchez, a Battelle Tech Fellow and John Snyder, CEO of NeuroLife. So. You mentioned that this is kind of like a part of PT, like physical therapy, right? So the patients are not taking the sleeve home, right. They're just, they're coming to you and using it for sessions, right, is that? Correct, that's correct. And so the video that I watched showed a patient who was trying to like stand up basically like a wooden cylinder. And in the first few tries, he couldn't do it. And then eventually he did it over and over again really well. And then the video showed that he was able to take that same movement and turn it into. Basically filling a cup of water from the refrigerator.
You know, the water thing. Yeah, yeah, exactly. OK Yes, I'm in front of two scientists calling it the water thingy, but you know, um, okay So tell me about that. Tell me about the process. So how long did it take for him? I know he's just one example But how long do you take for them to get from not being able to hold the cylinder to suddenly being able? To fill a cup of water.
Snyder: So usually it's a 36 session protocol. And we usually see response at around 24 sessions, but we usually go to 36, so three times a week for that. And that seems to be the protocol, the durability that's needed to rebuild that neuroplasticity. And.
Juravich: And so can he now do it at home without the sleeve? Correct, yes.
Snyder: Correct, yeah, so there's another video that we have that shows a stroke study patient. She's trying to take a ball bearing and place it from this small canister up onto this other canister. She's unable to do it. And this is at the baseline. And then we showed with that, after her treatment protocols, she was able to do in six seconds. And she also did not have the sleeve on when she did that, which shows that it's not an assist device, as you call it, or as some people will call it. It's truly a rehab device and initial market will be in that PTOT rehab setting for patients to come in and basically be rehabbed with the device there.
Juravich: And Justin, can you do this on the right side and left side?
Sanchez: Right and left side and here's a wonderful part about the nervous system. It's supposed to be engaged with your body in the world and the more that you can actively engage it, the more it can adapt to whatever situation and as we've been talking here, we've seen a lot of interesting examples of how people have used this technology for a while and then when they take it off, those benefits that they're starting to see are persisting.
Juravich: Are there differences between a stroke victim and someone who had a spinal cord injury with the way the brain talks to the arms? Like, do you have to do different things with the sleeve for what happened to the patient?
Sanchez: Yeah, so depending upon where that injury is, so strokes are typically in the brain or around parts of the brain. So we got to consider that there's a difference in the signal from there. Or in the case of a spinal cord injury, it could be, let's say, in the neck where your spinal cord is the neck and the signals may not get from the brain to the nerves and the muscles that are in your limb. And our system is flexible enough that it can adapt to either of those kinds the scenarios.
Juravich: And so, John, it sounds like you are planning on taking this broader. You want more people to be able to have access to it. Right now, is it just you have to come to these two places around Ohio State to do it, right?
Wan: Correct, right.
Juravich: Okay, so tell me about going bigger and cuz you mentioned it's gonna take a couple of years and no one ever likes to hear That so go ahead. Tell me how you're gonna go faster
Snyder: So basically, we'll launch a couple of larger studies in spinal cord injury and stroke, so chronic stroke. And we'll be at some leading institutions. So Ohio State will still be one, but we're talking about the leading institutions in the U.S. So we'll have a nice broad representation of it. But we basically have all of them nodding heads as far as saying, yeah, we want to be part of this. They're excited about this because of, as what Justin said. The sensing piece of us and the closed loop piece of this. There's really nothing out there that actually does this as far as the sensing and directing that stimulation during the rehab itself. So they really wanna be part of this next evolution of rehab technology. So things just take a little time. So you need to get specific approvals at these centers to move forward, clinical trial agreements, et cetera. We need to manufacture some more systems. We're talking to third party contract manufacturers to do that. But we target having this study up and running next year, and then we'll run it for about a year, roughly, and then submit to FDA for approval sometime in 28. So it's still pretty quick in our world. That's a quick, yeah, I understand. We wish it was there already, you know, tomorrow, but we just have to go through the right paces to get it through the process.
Juravich: Justin or is Battelle building them?
Snyder: Or just.
Juravich: No.
Sanchez: No, so Battelle's doing the early prototyping in the foundational technology development. And we work with John and his team in order to figure out the manufacturing and the next steps and the design parameters that might be needed for manufacturing.
Juravich: Okay. And then do you have to get like, do you have to patent the technology? I mean, is there anyone doing something similar to this?
Sanchez: Yes, so Battelle has many years of history of patenting every component of this technology that you could ever possibly imagine. Well, because you've been working on it for 10 years. For years, yeah. I believe we have over 100 patents in this. In the sleeve. In the sleeves and all the related technology, the algorithms, the sleeves, the hardware, all of those things that go along with it.
Juravich: I made it seem very simple the sleeve, but I mean, is each sleeve like a million dollar product or like what are we talking here?
Snyder: I don't want to share the price. It won't be a million dollars, but so we have a different targeted price for that Yeah, one of the things that we didn't talk about is when you're developing new technology It's so important to have reimbursement coverage for that because otherwise it won't get
Juravich: Yeah, I was just going to ask if
Snyder: Insurance covers it. OK, that was my next, yeah. So there's current insurance coverage. It's called CPT codes in the medical device world. All that means is that it covers certain devices. So this would fall into those current existing codes, which is great for this technology. So
Juravich: So it qualifies the same way other pieces of physical therapy equipment would?
Snyder: Correct, exactly. So basically once we launch this will be reimbursed for those centers and so they'll be able to utilize it and get reimbursed for it which will increase the uptake and also help to get this to get them more and more patients out there.
Juravich: Okay, and this is intended only to be like a physical therapy item that you use with supervision, like no one's ever going to take these home, right?
Snyder: Well, not this version. Gen 2 could be an at-home version. It's a little bit different as far as the reimbursement path on that. So it's different than current reimbursement. Also probably would have to work in a telehealth type of component so that somebody is using it at home. But that's a second generation, probably a second focus. But there's other opportunities there. But it's interesting.
Sanchez: But it's intriguing, right? Imagine a future in which you could do rehab at home, right, in this very kind of targeted specific rehab that might help you to get to a level that you never could before. We're very intrigued by that.
Juravich: Intrigued is a word, right? Yeah, excited as well. Excited. Yeah, so the technology in it though, Justin, do you have to program it per person? Or how do you figure out to get it to work for each individual person to be able to do what they're trying to do?
Sanchez: So just back to the early days, we had some very talented engineers that would make these kind of AI algorithms that would do this interpretation. It took a lot of hours to do that. Over time, over that 10 years, we've made these systems more automated. We made it such that you can build the experience inside of these models, and then a person can, let's say, grow that model over time so that it becomes more and more personalized. We've even seen a few early examples of if you build a model for one person, it might have some foundational components that apply to another person. So the more that we use this in a lot of different research environments, the more we are understanding which components are really gonna apply to different clinical settings and then maybe even beyond the clinic in the future too.
Juravich: Is this an example, since you've been working on this for 10 years, is this an example of like AI has been around for a long time. We just didn't all talk about it every day for the past two years.
Sanchez: I'm so glad you mentioned that. If we had this conversation all the time, it's exactly right. There are foundational components of the models that we use for interpreting a person's intent to move that are based in AI, but we didn't necessarily call it that back then, but it's very much a similar kind of technology.
Juravich: Okay. So if everyone is feeling a little like apprehensive about AI, you're going to say, but hey, look what we can do.
Sanchez: That's exactly right.
Juravich: Okay, so the Neurotech Institute is a non-profit biotechnology organization. Is that different than NeuroLife? What's the NeuroTech Institute?
Sanchez: The Neurotech Institute was formed out of a partnership between Ohio State University and Battelle, and the whole idea was to make Columbus and the surrounding regions in central Ohio a hub for neurotechnology. We had so much talent in the region coming from the foundational work the university was doing, the applied science and technology that Battelle was doing. We said let's join forces and create the conditions in which we can launch new companies out of the most promising technology.
Juravich: Okay, so that that launched NeuroLife is what is what happened. Okay, I was getting mixed up with the neuro word. So neurotech helped launch a company and Neurolife is a company. Okay. Got it. So So what's next? Do we just focus on the sleeve? Or are we going to do other parts of the body?
Snyder: I mean, as far as neurotech is concerned, or... I don't know, yeah, I don't know. Let's start with neuro life. So, I mean our focus, obviously, stroke and spinal cord injury, but there's other clinical indications where we can deliver this therapy in clinic. So, it could be Parkinson's, could be MS, could be spasticity. So, a variety of other areas, too, that patients need help on with regards to that rehab, and... This could be a great model, a great avenue for them to get there. So we see other opportunities there too. What's interesting though is, I mean, we'll have to do additional clinical studies in those areas, but we could build off the safety and efficacy of the studies that we've done so far and those future ones that we're gonna do in the coming year or two. Okay, yeah.
Sanchez: So from a company perspective, focus is just so important and John and his team, they have exquisite focus. Where Battelle comes in all this, we're always thinking over the horizon. So we're thinking about next generation technology that's not just on the limb, it could be on different parts of the torso or on the lower limb or the leg and could be applied to medical or even non-medical kinds of applications. So it's our role at Battelle to think about science and technology for the betterment of humanity and we're think about all kinds of application of this technology in the in the future.
Juravich: All right, well, this is a wonderful advance. When you get those FDA approvals, come back and tell me all about it, right? We've been talking about new rehabilitation technology for stroke victims and those with spinal cord injuries. And we've been with John Snyder, CEO of NeuroLife. Thanks for being here. Thank you. And Justin Sanchez, Battelle Tech Fellow. Thank you, Justin.
Sanchez: Thanks for having us.
Juravich: And this is Tech Tuesday. Coming up, we are going to talk about two of OpenAI's most capable models who hacked an AI startup. They did this unprompted with limited guardrails. And what happens now? That is when Tech Tuesday from All Sides continues on 89.7 NPR News. You're listening to Tech Tuesday from All Sides on 89.7 NPR News. This is a show where we share stories about science, technology, and the future of our environment. Two of OpenAI's most capable models hacked into an AI startup. The AI did this unprompted with limited guardrails and it's raising some concerns. Also, Samsung recently unveiled some new phones with promising features, including some smartphone origami. And we are gonna talk about people reverting to analog media, Polaroid cameras are making a comeback. To tell us about this and more, we have Russell Holly, Director of Commerce, Content for CNET. Welcome back, Russell.
Russell Holly: Thank you.
Juravich: So I want to start with OpenAI. So two of OpenAI's most capable models hacked into what's called Hugging Face, which is an AI startup. And the OpenAI models did this on its own. It was not prompted. It has a lot of people worried. So first off, let's start with what is Hugging face that it hacked into.
Holly: Hugging Face is a different company that functions kind of within the AI space, and it serves a couple of different purposes. Big thing is community-building, so being a place where you can kind of compare how different AI models behave and things that they're able to kind of do together to choose what the best tool would be for whatever job it is that you're trying to accomplish. That's that's kind of the. A very large, you know, kind of overarching explanation of what hugging face as an organization is all about.
Juravich: So why did open AI hack into it?
Holly: That is a really good question. I want to start all of this by saying that there is a staggering amount that we do not know about this situation. OpenAI has a history of not being particularly transparent. Hugging Face is able to make some specific claims about what it received as far as the hack and what information was being accessed. But OpenAI, as a company, has not released an enormous amount of fact-based information. There's a lot of guessing that's going on here. But to start, there is a version of ChatGPT called 5.6 Sol that has not been publicly released yet. It was something that was being tested in what OpenAI calls a sandbox, which is a computer that is removed from everything else so that it can perform some tests and you can measure what that AI tool is able to do, you know, with in kind of an artificial environment. The software here located a way to explore things outside of that sandbox and located a ways to access the Internet through exploits that are not immediately clear. There are some claims that these were what are called zero-day exploits, which means exploits security researchers were unaware of that were discovered. But access the internet in order to locate the hugging face servers and penetrate those servers and go around looking for information. According to the people at Hugging Face, this software was able to kind of rummage around in their data for a full week before someone at OpenAI or Hugging face knew that this was happening.
Juravich: Oh, wow. Okay.
Holly: And it's not immediately clear what the results of that were.
Juravich: Yeah. So, and is open AI saying that a human did not prompt it to do this?
Holly: It is correct that there was not a specific prompt for a human to do this. However, there is some really interesting language about whether or not GPT Sol made what OpenAI calls inferences based on conversations that a human being had with the chat bot.
Holly: Because remember that all chatbots, regardless of what their emergent tools are, they record conversations that have been had and have been known on many occasions to pull information from previous conversations in order to apply context to future situations. Not in a thinking way, but in accessing prior bits of information and trying to fit those two things together. So while it is likely true that someone who works for OpenAI did not specifically give the command to break into Hugging Face and get some information, there is suspicion that folks in OpenAI had been talking about how the people at Hugging had information. That were needed to cheat some sort of test, and that that was the information that this version of ChatGPT was looking for.
Juravich: So I mentioned the word worry, I mentioned concerns. I mean, how close are we getting to an uh-oh moment with AI?
Holly: I mean, this is very much an oh-oh moment with OpenAI. It's also very far from the first one. OpenAI has had a number of internal breaches where its software has not done exactly what it was that it was told to do, where different kind of logic patterns were found in order to make different decisions. And there is just not a lot of transparency from OpenAI about how these things were located, how they were found, what was done in order to stop these things from happening. A lot of that comes back to just the way that OpenAI as a company tends to operate. This is very much an organization built on the Silicon Valley model of moving fast and breaking things, and they are very good at moving fast, and apparently also very good at breaking things.
Juravich: Will we ever know really what happens or do we not have the like open AI doesn't have to tell us?
Holly: OpenAI isn't short of a government investigation or some sort of lawsuit from the people at Hugging Face. It's not immediately clear how the company is going to respond to this. It doesn't seem like there was a massive financial kind of reason for OpenAI to be super clear about the information that's going on here. But it does raise some significant questions about whether... You know organizations in you know kind of corporate environments or or you know fortune 500 companies like you are you are investing heavily in this organization that is not being entirely forthright with you about what its capabilities are and that That should alarm anyone running those companies
Juravich: Duly noted. This is Tech Tuesday from All Sides on 89.7 NPR News. We're talking about some recent tech news with Russell Holly, Director of Commerce Content at CNET. I wanted to move on to Samsung. They recently revealed some new phones with promising new features, including better battery life. They do some smartphone origami, lots of folding happening here. Traditional smartphones are very slim. They're vertical. You hold them like a rectangle that you hold vertically. How do these foldable phones change the aspect ratio? So why are these new Samsung ones desirable? Because they're kind of like bigger and squarer and the rectangle's different. I don't even know how to describe it. Yeah, go ahead.
Holly: It very much depends on which version of the phone. So for the last couple of years, Samsung has been releasing two different folding phones, what they call the Z Flip, which flips kind of the way that your old Motorola Razr would flip and is very kind of small and thin and candy bar style. And then they have what they called the Z Fold, which folds out like a book. And gives you a smaller screen on the front, and then you can kind of open it up and you get a wider screen. And there has been some debate among other companies that make versions of these folding phones about what the perfect aspect ratio for these things are. There is, Samsung for quite some time wanted to make sure that when the phone was closed, it still had the rough shape of a kind of candy bar shaped phone. And could still very much fit in your pocket, but it turns out that a lot of the folks who are a big fan of these wider folding phones want the biggest possible screen that they can get. And so to address that audience, Samsung announced the Z Fold Wide. And as the name suggests, when you open it up, it is very big. When it is closed, it really more resembles a deck of cards than a traditional phone as far as its kind of general shape.
Juravich: And then it opens up to be the size of like a tablet, you know, like an iPad to talk about the other company.
Holly: Yeah, it is a 4 by 3 aspect ratio, which is something that is unusual when it comes to phones and tablets, in that it doesn't have the kind of widescreen presentation for consuming video. And we'll leave kind of black bars on the side, which is something that for a very long time companies were kind of keen to avoid. But yeah, this is very much aimed at giving someone the largest possible screen that they can because there is a group of people who own these phones who will use them as their only traveling computer, right? If they're in a hotel room, they'll bring a keyboard with them so that they kind of type in a bigger fashion. And with that comes a need for a phone like that to kind of have a battery that is big enough to function also as your laptop, which is a big part of what Samsung was promising here with this eighth generation of these folding phones.
Juravich: Yeah, so the there's what is it called a silicon carbon battery? So much more much more battery life, right?
Holly: Yeah, this is actually really interesting new battery tech that allows it to work better with kind of a phone that is folding without needing to compromise where the battery lives inside of the hardware and is something that could prove to be a new standard for phones moving forward because of the increased battery capacity that lasts in that. It is worth pointing out that Samsung has had some battery problems in the past with phones that have been recalled because they were exploding. Which is not something that is a concern for these phones, but the idea of Samsung playing with brand new battery tech is something that gives some consumers pause due to the history associated there.
Juravich: Are they heavy?
Holly: They are, obviously, because they are twice the amount of hardware as a normal phone. They are noticeably heavier than the iPhone you currently have in your pocket, but not in a way that I would say kind of drags around. If you regularly have your phone in a bag, I wouldn't say that you would notice any difference. You're not putting these in your back pocket, regardless of the pocket situation on your pants. But yes, you will notice that these are kind of larger and heavier phones.
Juravich: We have talked about foldable phones many times on this show. I don't have one, but I watch an unboxing video and I'm still blown away by the fact that it doesn't have any creases. I mean, how do they do that? Does it crease over time? I just don't understand.
Holly: It definitely creases over time, but it has been, this is something that Samsung is now on their eighth generation of this. So they've been doing this for eight years now and they were not the first company to do this. So we're rapidly approaching a full decade of companies trying to make sure that that seam is as unnoticeable as possible. So when you first get a Samsung folding phone, when you open it for the first time, you will notice that there is no visible crease there. But due to friction and use and things like that, over time that crease does appear and it does become more noticeable the longer you use the phone. I know many people who have had older versions of these phones and right around the two year mark that crease becomes very noticeable. Hmm. Okay.
Juravich: And just to end on with the foldable phones, when are we gonna be talking about iPhone having one? Because are they eight years behind? Is that what's happening?
Holly: That has been a question for a very long time. Apple is not a company that prefers to be first at anything. They would prefer to be right the first time that they release something. And so, I wouldn't expect one this year, but I also wouldn't be surprised if it was announced as something for next year, kind of a one more thing, which Apple occasionally does, especially since this year will be the first year that Apple's brand new CEO will be the one delivering the iPhone presentation in September.
Juravich: All right. Well, for our one more thing, we have one more thing to talk about. We have Polaroid cameras are making a comeback. So instead of all of your photos just sitting in the cloud, what's better than holding a picture in your hand? Tell me about this Polaroids GoGen3. What are some of its features? Do the pictures look good? What's going on with the Polaroide camera?
Holly: The most important feature, honestly, is that it is adorable. This is a cute little piece of rounded plastic that has great big lenses on the front. It does not look exactly like the Polaroid camera of your childhood, but it certainly resembles it. It has a great big flash on the front and a kind of concave mirror so that if you go to, you know, the kind of classic selfie with a Polaroids, because that was a thing we did before phones, no matter what your kids will tell you, Uh, and the- The whole, you know, kind of deal here is that it can function exactly like a Polaroid. The image quality is not going to be the same quality that you would get from your phone. However, you can also turn on Bluetooth for this Polaroide Go Gen 3 and send pictures from your phone to it to be printed. So if somebody else has a picture that they really like or a picture you have taken that you really like and it wasn't taken on this Polaroid but you have it nearby, it can just be a photo printer.
Juravich: Okay, so it's a Polaroid camera and it's a printer. Wow. Okay.
Holly: Yeah, because if you use it just as a Polaroid, it's the exact same deal. You press the button, you see the flash, and then the picture comes out a couple of seconds later and slowly comes into focus as you kind of wave it back and forth. That same process is very much true here.
Juravich: Do we have to shake it?
Holly: You don't actually have to shake it, but I'll tell you, we used it several times and I've yet to see a person not shake it the moment that they hold it. There is some sort of muscle memory there that millennials in particular just can't get rid of.
Juravich: You just feel compelled to shake it, but you don't need to. You can just leave it sitting there. Because I've read many things that say that shaking it doesn't make the picture appear any faster, right?
Holly: No, that's exactly right, but I feel like it's just something that you can't help yourself but do.
Juravich: Okay, well how much does one of these GoGen3 cameras cost?
Holly: These are $90 to start, and they come in two different colors, but the one that is the most common is that kind of flat Polaroid white. Do you know how much the film? With a great big orange. How much?
Juravich: How much does the film cost?
Holly: Bill, cartridges are going to run you about $20, and they come with a, you know, you can buy them in kind of larger packs, but each one of those cartridges are going be about 120 pictures.
Juravich: Oh, that's a lot. Okay, right. Yeah. All right. My full the Polaroid camera of my youth was like 24 pictures or something like that, right?
Holly: If you were lucky, yeah, I remember the 12-picture packs being very frustrating for a while there.
Juravich: All right. All right, we've been talking about foldable phones and Polaroid pictures. And then earlier we talked about AI going rogue with Russell Holly, director of commerce content for CNET. Thanks so much for joining us, Russell.
Holly: Thanks for having me.
Juravich: This has been Tech Tuesday from All Sides on 89.7 NPR News.