Drones and Robots are Becoming Common Food Deliverers - DTNS 5333

0:01Back to school is busy enough. The last thing you need is shopping that makes it harder. Whether you're moving into a dorm, your first apartment, or just getting organized for the semester, Wayfair makes it easy to get everything you need in one place. Furniture, bedding, storage, decor, study essentials, all at prices you'll love with fast shipping before move-in day. Get everything you need for back-to-school season. Shopwayfair.com today. way fair every style every home this episode is sponsored by ZocDoc have you been putting off a doctor's appointment or need a dentist and you don't know where to start we've definitely been there and that's why we use ZocDoc actually over the holidays Michael had a dental emergency and seemingly every office was closed but we hopped onto ZocDoc found an in-network dentist had great reviews, saw real-time appointments available, which was key.

0:56It was on New Year's Day and booked an appointment in just minutes. It took so much of the stress out of the already stressful situation. ZocDoc makes it easy to search and compare over 250 ,000 local in-network providers across more than 200 specialties. So you can find the right doctor or dentist and book instantly. Taking care of your health just got easier. Download the ZocDoc app today and book that appointment on your to-do list. That's Z-O-C-D-O-C. ZocDoc.

1:30Dr Niki: This episode is sponsored by HeyGen. AI is great at saving time, but only if it actually fits into your workflow. That's what makes HeyGen so compelling. Instead of spending hours setting up lights, cameras, and retakes, you can turn a script, photo, or presentation into a polished video of you in minutes. No production crew required. Whether you're a creator, consultant, attorney, financial advisor, or real estate agent, HeyGen helps you create consistent, professional video content without making content creation your full-time job. It's rated the number one AI video platform for small business on G2, features some of the industry's most realistic AI avatars, is trusted by more than 30 million people and 85 % of the Fortune 100, and supports over 175 languages.

2:12Dr Niki: So you can communicate with audiences around the world. Here's the wild part. Record yourself for just 15 seconds, and HeyGen builds an AI avatar that makes professional videos for you on demand, so you can post everywhere your audience is, as yourself, without filming every time. Your first three videos are free at heygen.com slash pod. That's H-E-Y-G-E-N dot com slash P-O-D. This episode is sponsored by Morning Brew Daily. If you're anything like me, you want to stay informed about what's happening in business and finance, but you also don't have an hour every morning to sift through headlines.

2:48Morning Brew Daily is the go-to business news podcast for millions of people. It's a daily business news podcast that's a quick, fun, and easy way to start your day. Whether it's the latest in tech, markets, or the trends shaping the economy, the show breaks it all down in a way that's clear, engaging, and actually enjoyable to listen to. Hosts Neil Freiman and Toby Howell make the show feel like you're catching up with two friends while giving you the context you need to feel smarter and more confident throughout your day. New episodes drop every weekday wherever you get your podcasts. Tune into Morning Brew Daily wherever you get your podcasts.

3:30Dr Niki: This is the Daily Tech News for Monday, August 17th, 2026. We tell you what you need to know, give you some important context and help each other understand. Today, Dr. Nikki follows up on the combo of AI and dogs to fight cancer, and drones and robots expand your food delivery options. AI dogs and robots, what's not to love? I'm Tom Merritt. And I'm Rob Dunwood. Let's start with what you need to know with that big story. So yeah, Uber is partnering with Zipline. You may recall longtime listeners of DTNS may recall that Zipline kind of made its name delivering medical supplies in Rwanda. They have operations throughout the world now, including in the United States.

4:11Dr Niki: So Uber is going to partner with Zipline to begin drone delivery of meals for Uber Eats this year. If you were worried that Zipline was going to do a drone delivery of people, it's Uber Eats. It's Uber Eats going to be delivering food. The partnership will start in Zipline's existing U.S. markets. That includes Pea Ridge, Arkansas and Dallas-Fort Worth later this year before expanding to a few dozen more cities. Zipline currently delivers for Walmart. That's probably why Pea Ridge, Arkansas is on there. And also for Chipotle in the U.S. So they're adding Uber Eats to the roster. Uber says it hopes deliveries should be able to be completed in 5 to 10 minutes.

4:49Dr Niki: Now, I assume that's 5 to 10 minutes from the time the meal is completed. But still, much faster than the 15 minutes maybe the driver would take to get there. Uber says it has a goal of reaching 1 million drone deliveries a day by the end of 2029. Much more ambitious than what we've seen with others. Uber has also previously partnered with Flytrex, still has a partnership with Flytrex. Flytrex is also working with Walmart. So Walmart on the forefront of this. DoorDash and Wing recently partnered and expanded their delivery service to include the Atlanta metro area. And meanwhile, on the ground, former Uber company Serve Robotics has signed up Grubhub to use its sidewalk robots to deliver food in Alexandria, Virginia, Chicago and Los Angeles.

5:34Dr Niki: Serve Robotics already delivers for Uber Eats as well as DoorDash. Rob, we are we are getting to the point where these are in the thousands, you know, and Uber wants to make it millions a day. adding a few more delivery areas here and there. I see these robots around Los Angeles. How about you? So where I'm at in the Midwest, we don't see them that regularly, but when I visited large cities that have them, I'm kind of like, oh, that's why these will work. These areas that are very, very densely populated, particularly for the robots that are on the ground that are delivering stuff, I can see it because it's like somebody's got to walk or ride a bike to do the delivery.

6:16A robot could just as effectively do that maybe in some cases more effectively do it. So I think that this is just going to be the way of the future. And then we'll go back to the drones actually flying through the air to deliver you stuff. It's not even really the last miles, like that last, you know, 1 ,500 feet, that last, you know, 2 ,500 feet or however far they've got to go from their base station is where all the time takes with delivery models. I remember being at a swim party in a very, very large complex, And there's just multiple UPS trucks, multiple Amazon delivery trucks. There's just multiple trucks just driving through the neighborhood for literally hours at a time.

6:55And it's just because they got to, you know, this was on this truck, that was on that truck. But if they could have that stuff all delivered individually via a ground-based robot or a drone, it would save them a lot of time, a lot of manpower. As far as these things, you know, being done by things that are not manpowered. They're powered by batteries and they fly through the air roll on the ground.

7:15Dr Niki: Yeah. This is a low margin business because it only works if it reduces the cost for the clients. But it is reducing the cost of these deliveries. It's speeding them up. Uber Eats saying it wants to do five to ten minutes from the time that pizza is cooked or that burrito is made. And they're complimentary. I actually didn't think about this until you were talking right now. Yeah, you see the sidewalk robots in denser areas. In fact, I don't even have them yet in my area of Los Angeles. They're more in the more dense parts of Los Angeles. But the tag on the drones has been, well, these only work in the more open areas, in like rural areas where you need to fly farther than a sidewalk robot could materially speed up and make a delivery.

8:02Dr Niki: So I think if you've got drones on the edges, you know, your suburban and rural areas and you've got sidewalk robots in the dense city centers, suddenly you've got some decent coverage going on. Yeah. And the thing with the ground-based delivery robots from Serve, it looks like they're just trying to be, it doesn't matter if it's Uber Eats, it doesn't matter if it's whomever. Whoever has stuff that wants to be delivered, we're your company. It looks like that's kind of where they're building out. They want to be the company that you go to when you've got stuff that needs to be delivered on the ground via a robot.

8:35That's just it seems like they're setting up their infrastructure to be that kind of player. Yeah.

8:40Dr Niki: Sidewalk robots as a service or SRAs, I will say. I would just point that. Well, yeah, I think I think we're still I'm very curious if they can get to a million deliveries a day by 2029. But, you know, we might be a couple of years out from this being common in major cities, maybe even in Columbus. You know, like you're big enough that this should come there. Yeah, I'd love to see him here, but it'll depend on how far out in the suburbs they get. I can see it downtown. Yeah, yeah. It'll get here to the suburbs and sometimes. I mean, it's not even in the suburbs here. DG Ness is made possible by you, the listener.

9:22Thanks to Dale McKay, Matt Zaglin, Jeff Wilkes, and Todd Carozzi.

9:27Dr Niki: Yeah. And we got a couple of new patrons, Chris and Charles, as well as a raise from a different Chris. Love all the Chrises this week and the Charleses. And returning patron, Surfer, no vowels, S-R-F-R. Welcome back, Surfer. Good to have you. episode prep gets messy when research slack threads and script trackers are all in different places that's where chat gpt work comes in give it the goal and connect to the tools where work happens like apps and files and it can help pull the pieces into a draft you can review without you having to re-explain the project at every step for example chat gpt work can gather team feedback from Slack and use it to suggest updates to a project brief in Google Docs.

10:14You stay in control reviewing the draft and deciding what moves forward. Put ChatGPT to work on your most ambitious ideas and projects. Get started at ChatGPT.com by selecting work mode. Available on Plus and Pro plans. Standard size rugs shouldn't be the standard. Get a custom sized rug that actually fits with Ernesta. In as little as two weeks, Ernesta cuts, finishes, and delivers your rug straight to your door. No more awkward sizing. Just a designer-quality rug made to work within your space perfectly the first time. Visit Ernesta.com to explore Ernesta's full range of custom-sized rugs. And get$50 towards any style when you order five samples for only$10.

11:00This episode is sponsored by ZocDoc. Have you been putting off a doctor's appointment or need a dentist and you don't know where to start? We've definitely been there and that's why we use ZocDoc. Actually, over the holidays, Michael had a dental emergency and seemingly every office was closed. But we hopped onto ZocDoc, found an in-network dentist, had great reviews, saw real-time appointments available, which was key. It was on New Year's Day and booked an appointment in just minutes. It took so much of the stress out of the already stressful situation. ZocDoc makes it easy to search and compare over 250 ,000 local in-network providers across more than 200 specialties.

11:43So you can find the right doctor or dentist and book instantly. Taking care of your health just got easier. Download the ZocDoc app today and book that appointment on your to-do list. That's Z-O-C-D-O-C. ZocDoc.

11:58Dr Niki: In a world where AI is changing every role in every industry, knowing who actually has deep expertise matters more than ever. Upwork connects you with the highly skilled freelance professionals with proven track records, so you can hire with confidence. Upwork gives your business fast access to highly skilled specialists with deep expertise, so you can fill critical skill gaps, launch projects faster, and scale support up or down without committing to full-time headcount. With Business Plus, you can access the top 1 % of talent on Upwork. AI-powered shortlisting delivers a curated shortlist of top freelancers matched to your goals in under six hours.

12:34Dr Niki: No endless searching required. Thousands of growing businesses already trust Upwork to hire highly skilled freelance professionals for everything from one-off projects to ongoing support. It's free to sign up, and posting a job is easy. Visit Upwork.com right now and post your job for free. That is Upwork.com to connect with top talent ready to help your business grow. That's U-P-W-O-R-K dot com, Upwork dot com. Hi, this is Kim and Penn Holderness from the Laugh Lines podcast. This episode is sponsored by Oli, your wellness partner for back to school and beyond. Back to school season is here. Packed backpacks, packed lunches and packed schedules are officially back.

13:13Dr Niki: Oli Kids Sleep helps little dreamers wind down with a gentle blend of melatonin, L-theanine and botanicals for restful sleep. Translation, an easy way to put those school night negotiations to bed. And for the grownups, Ollie's Sleep helps you fall asleep, stay asleep, and wake up feeling refreshed. Because as you know, the morning mayhem comes early. Visit Ollie.com and stock up on back-to-school must-haves today. That's O-L-L-Y dot com. These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.

13:51Dr Niki: There's more we need to know today. Let's get to the briefs.

14:21If no one ever leases the data center and the building is sold, even then we'd only have to pay the difference between the sale price and its guaranteed amount.

14:29Dr Niki: And there's even a clause in there that says if they end up having to pay something, OpenAI will pay them back. They'll indemnify them, which I'm like, well, if OpenAI is so insolvent that they can't do this lease, I don't imagine you're going to get your money in video. But yeah, this is a much safer way to do it to say like, hey, we'll back. We'll make sure it's basically like a bridge loan. We'll make sure that it's worth building this building because once it gets built, it's probably going to be worth$1.5 billion, which is what NVIDIA is guaranteeing here. Plus, NVIDIA also gets to sell a bunch of chips here.

15:01Dr Niki: I think the more important thing is closed loop and air cooled. That is very good, especially at a data center this size. I think it's going to be 4.25 gigawatts off the bat, but with a goal of eventually building out 10 gigawatt capacity. So you want that to not be using your water, and it won't. However, I don't like natural gas as the energy, which is becoming more common for these things. Yeah, it's like they're trying to make them as clean energy as they can. But if you're using natural gas, if you're using anything other than wind and electricity, they're not completely clean. So it would be nice if they could get to that point.

15:39But this is a big data center. And from NVIDIA's standpoint, it's like, you know what? We'll back them. We're going to make the money back. This is almost like being like a land man or something. I've been watching land man lately. Oh yeah. Just watching how it's like, so we really don't, we, we never lose. We never lose the buddies. Like we'll buy this stuff. If it doesn't work out, we'll just sell it to somebody else. That's kind of what they're doing here. It's like, we'll cover it. If it doesn't work out, it'll get sold. It'll probably get sold for more than we, than we need to cover it for.

16:07So, so we're good at the end.

16:09Dr Niki: Yeah. They're literally land man. Cause it, you know, data center is just a big local area network. Right. Yeah, no, I think that's smart on NVIDIA. It certainly is causing a lot of gnashing of teeth, the amount of debt these companies are going in to build this stuff. But I think that gnashing of teeth is mostly around the people investing in the companies building it, because the buildings themselves will be used for something. I feel pretty confident about that. And NVIDIA feels pretty confident about that. Anthropic has published a layman's explanation of its text watermarking, which makes it clear what's going on.

16:48Dr Niki: It's something that I was hoping they would do, and I'm glad they did it. Anthropic makes the point in their explanation that the watermark provides a signal that the content was processed by Claude. It does not prove that the text was generated by Claude and should not be used to detect LLM-generated text. This is a concern I saw a lot in our chats and emails and such. uh anthropic is saying if you put your own handwritten text through here for for for processing to like bold stuff this watermark will not show that you didn't write it likewise if there's no detectable watermark that doesn't mean that the text was not generated uh there are ways to get rid of this watermark but it's fairly durable dclad.org wrote a very good explanation about how the text watermarking works including some fun sliders that'll let you see how editing the text will affect the watermark.

17:41Dr Niki: Basically, a full rewrite will get rid of the watermark. But the key to the watermarking is that it is tilting the probability of word choice. It's basically looking at each word that's about to be detected, and it then creates a list of green and red words. The red words are less likely to be picked. The green words are more likely. Red words can still be picked, so it's not determinative. It's not going to fully change the text, and it's a different list every time. So if you are the key holder, you'll be able to look and see how that text was generated and whether the watermark is there. But teachers and independent detectors will not be able to use this to detect AI composition unless they have the key.

18:28Dr Niki: And Anthropic says we're the only ones that will have the key. So we will run a watermark detection tool for people to use. But without the key, this is not something you can easily just look at the text or even scan the text and figure out what the watermark is, because it's not always the same word choice that's being tilted. And they reemphasize a watermark is not an AI detector. It means the text was processed by Claude, not written by Claude. But Rob, I'm a little disappointed with this because it works kind of the way the Synth IDU does for Google, which has the chance of changing your writing, even if you tell it not to because it's using the watermark.

19:04There's that. And you actually explain this. You're pretty dead on last week. You know, you had a good idea of how this is going to work. The interesting thing to me is that because this is not an indication that it was AI, it's just this is an indication that we may have or may not have actually done something to your code as it comes through us. That is that is kind of interesting. So I'm just thinking it's like, so so is this AI? It could be. Could not be. Did you guys modify it? Maybe. Maybe we did. It's like, you know, who knows except for anthropics. So it's kind of funny to be in that sense.

19:37But it is interesting. They're saying that it won't be easily detectable. I wonder how easily detectable it will be. Because I hear people saying challenge accepted. I will try to figure out. We will try to figure out how this works.

19:52Dr Niki: And SynthID by Google is a good test case of that, which it has proved not to be that easily detectable. Not easy at all. Not easy at all. So students who are going to use AI to write the term paper, they have let their shoulders down and they've taken a sigh that they're not going to get caught by. You still shouldn't be doing it, though, students. You still shouldn't be doing it. There are other ways to detect it that they could fall subject to. Anthropics just saying, don't use this. There are other tools to do it. Don't use ours to do it. And our watermarking is not going to make it easier for those third parties necessarily.

20:26Dr Niki: Because if you look into the explanations, it is really interesting how it is a very difficult pattern that it's putting in there that wouldn't be obvious to detect. And like I said, there were two ways to doing it. I was hoping it would be in invisible characters because, yes, those are easier to strip, but it won't affect my actual text. And it sounds like it won't. On the other hand, if you're a coder and you're like, wait, is it going to change my code? No. It actually is not very robust when it's generating code at watermarking because it has restrictions of like, we don't want to change code.

21:03Dr Niki: It's also not very good at short versions of text. The longer the text you're putting in, the better of a chance you have to get a good watermark out of it. Yeah. And you kind of alluded to this earlier, but as an author, you can't be terribly happy about this because if you're using it to not write your content, but just let's go in here and clean up some stuff. Where do I have, you know, hanging part of stuff? But, you know, help me do what AI is good at doing. It now maybe changes what you've written. So you've got to, there's got to be a way for you to actually stop that. It's like, no, I want what I've written to be what I've written.

21:33I don't want any words changed, regardless of what you're doing on the back end to help me correct it. So I hope that there's going to be like an undo button or something that they have for folks like yourself that are authors that are writing, that they want the absolute words that they've written down on the paper to be the words that come out for somebody to see.

21:49Dr Niki: Well, like the writing we do for Daily Tech News Show as well, not just for my books, right? Like I don't need to change what I wrote. I pick these words very carefully. So yeah, Gruber at Daring Fireball was going off about this. He is very upset that this is what Anthropic is doing. He's very upset that this is how Google does it as well. I have not, in using these tools to do like formatting and bolding and fixing commas and stuff, found that they end up changing the text all that often. So I think it may be a little bit of an overreaction. But the possibility is there. And I don't like it either.

22:26Dr Niki: I'm with them that far. Well, folks, we do live streams on YouTube. If you want to watch us on Thursdays, we record DTNS live. You can catch that live. And we do live streams of reading your emails and doing office hours. We have a tech tournament. It's the tech tournament of storage devices that is coming up. You can catch all those by becoming a subscriber at youtube.com slash daily tech news show.

23:01This message is brought to you by today's sponsor, Walden University. Set a course for change. Now is the time to make a real change in your life, career, and community. At Walden University, we give you the support and flexibility to get the W. Walden's online programs are designed for working professionals like you. You'll gain hands-on skills to take on real-world challenges and succeed. Your future is waiting, and Walden is here to help you achieve it. Take the first step to make your impact. Visit waldenu.edu. Walden University. Set a course for change. Certified to operate by Shev. Hi, this is Debbie Millman from the Design Matters podcast.

23:41and today's episode is sponsored by Lexus. Design isn't just about function. It is about feeling. At Lexus, every detail is intentionally designed. Master artisans don't just focus on what you can see. They obsess over what you can feel, the way the door closes, the precise click of a switch, the acoustic feedback of every control. Because a car that doesn't make you feel something is a car that stops short of amazing. Experience amazing at your Lexus dealer.

24:20Dr Niki: Let's check out some quick headlines that are going to make you look smart because you know these. Payment company Stripe has agreed to buy OpenRouter, a company that routes requests across more than 400 large language models. OpenRouter has called itself Stripe for AI. I probably would have called it Zapier for AI, but since they own Stripe or Stripe owns them now, I guess it makes sense. I guess, yeah, Zapier also works, but this is more about the like, you only have to pay once. It makes it easy to pay and then you choose the model. And Stripe must have liked that they called it that because they bought the company.

24:51Dr Niki: Apple has agreed to change its app tracking transparency system. That's the one that says like, hey, this company would like to have your data. Would you like to give it to them? They are changing it to meet requirements from regulators in Germany. Apple requests for your content to provide data to third parties will now, at least in Germany and the United and the European Union, remove possibly discouraging symbols and warnings, make prompts neutral in content wording and layout, and allow third parties to explain why they are requesting access to your personal data. The idea is to bring these more in line with the third-party data disclosures that Apple does for its own apps.

25:32Dr Niki: Germany is saying you are favoring your own apps by having stricter rules on this for third parties. LG signed an agreement with NVIDIA to use NVIDIA's Jetson Thor processor, Isaac Groot Foundation model, and Halo Safety System and bipedal humanoid robots to arrive in Q1. I bet we get to meet that at CES, don't you? I cannot wait. ByteDance and the Motion Picture Association have signed an agreement to strengthen copyright safeguards on ByteDance's video and image generation models, which means ByteDance can now sell their stuff to Hollywood companies, which they probably weren't going to be able to do without this.

26:17alibaba launched a beta version of a music generation model that can create a complete song from a line of text the model is called happy shrimp 1.0 and linus torvalds linus torvalds

26:30Dr Niki: announced yeah yeah uh the release of linux kernel version 7.2 complaining about the new normal of generated code adding more volume to a release than he would prefer in the final week uh he was It's like this was a bigger final week than I would like, but that's the new normal because everybody's generating so much code. The new kernel will support the Zaname Levelless controller, big esports thing, handles caches across many core chips, and makes it possible to run Linux on an Apple M3. The Wall Street Journal reports that Sony's CEO says that the company has not fixed a date for the release of the Sony PlayStation 6, and my gut tells me they don't want to release a date until they absolutely have to in hopes that the prices will come down.

27:13Dr Niki: I thought the Digital Trends article in this was really interesting because, yes, the price of the components keeps going up and they want to figure that out. But also they pointed out that all these AI tools that are driving up the price of the components are also being used by Sony's developers to make the PlayStation 6 better. And so they may be like, well, while we wait to figure out the price of the components, maybe we can actually have a better PlayStation 6 if we give these developers a few more minutes to work on it. Better and less expensive. Yay for both. I would like that. The Financial Times analyzed power commitments for 60 of the largest planned data centers in the U.S.

27:53Dr Niki: and found about 75 % of them are building natural gas facilities. So it's not just the OpenAI one. The rest are from renewable sources or from nuclear facilities. If they are all built with the current power sources when fully operational, this would contribute 7 % of the U.S. power sector's carbon dioxide emissions. Reviews for the new HP Omnibook X laptop are out with praise for the keyboard, trackpad, and OLED screen. The HP Omnibook X starts at$1 ,400 but is currently on sale at a discount. All right, those are the essentials for today. Let's dive a little deeper. An Indian startup called Dog Nasis is looking to combine sniffer dogs with machine learning to detect cancer from a single breath.

28:38So we had to get Dr. Nikki to look into it for us.

28:41Dr Niki: Yeah, last Friday, Jason and Huen talked about the company that is combining machine learning and dogs to try to help detect and perhaps treat cancer. So of course, we had to get Dr. Nikki to look into this. Thank you, Dr. Nikki, for joining us. And here I am excited to smell. I don't know. Sniff out the truth. Sniff out the truth on this story. Thank you, Tom. Yes. So, okay, let's get right to it. How do they plan on doing this? Okay. I saw this headline and I was like, I don't know, Tom, this seems pretty far-fetched. Ah! Fetched! Throw me a bone and explain it to me. It's actually really cool.

29:21So how it works. The patient breathes into like a cotton mask. They then ship this mask off to the Dognosis facility, which is the name of the company. It's not another bad pun. And they have trained beagles, Labradors, and Dutch shepherds that sniff this sample. Pretty straightforward so far. And here's where the quote-unquote artificial intelligence comes in. It's more machine learning. But as the dogs are sniffing, there's multiple sensors in this little workstation that they have that collect data on video tracking of the dog's movement, their breathing frequency their body language and their brain activity and they combine all of this into kind of translatable robust data that'll give you a yes or a no on the diagnosis and the founder says that this method targets over 20 types of cancer including breast lung oral and cervical and it looks like it has a promising future so i'm actually really excited about this

30:13Dr Niki: that is crazy i mean i've actually heard about dog's ability to to smell indicators of cancer Two things about this intrigued me, which is why I brought it to your attention. One was, of course, the machine learning aspect of it. But also, I'm glad you explained this to me. I was imagining when they said people breathed into a mask that like the mask was a little balloon of their breath. It just leaves the compounds on the mask. I honestly thought they like breathe into the dog. Hey, buddy, how are you doing? All right. So let's go further back, though. How does your breath contain anything the dog can smell to indicate that it's cancer?

30:59Dr Niki: How does that work? Yeah. So diseases like cancer, and this works for other diseases, I think they're just focusing on cancer, but they can cause subtle changes in your body's physiology, which then causes the volatile organic compounds or VOCs that your body emits to change. So your smell will change, whether humans can detect it, probably not. But dogs have a really, really strong sense of smell, as we know, and it's powerful enough to notice these subtle changes. And you can specifically train dogs who are good at this to detect specific VOC odor profiles, including 20 different kinds of cancer.

31:34Dr Niki: Wow. And I know there are other diseases that are detectable. I think I remember something about even COVID being detectable by smell. Yeah, apparently there's about 40 kinds of diseases that you can train dogs for. All right. So let's talk about how do they get the dogs to do this? How do they measure the dogs? I guess that's the really interesting thing for me is like, well, sure, you can train the dog to smell. And then you look at the dog and you say, is it acting like it smelled it or not? But this is making it more precise, I guess. Yeah, this is really cool. And actually, I was telling you earlier that I'm really excited that maybe I'll integrate this into my research because we do something similar.

32:12So first of all, I'm in awe of their system. Like I'm nerding out as a scientist. Like I don't know what kind of money they got, but or if they're just really smart, but it looks really cool. So each dog has a little workstation with like a little bowl with its sample of the cloth mask in it. each workstation has synchronized video and infrared sensors so when the dog gets in it triggers kind of a synchronization this is all synced up with a raspberry pi which again like the stuff you tinker with at home scientists are also tinkering with and then each dog has a really cute little sensor suit and a hat

32:45Dr Niki: it looks like a super dog an EEG cap so like to measure brain waves mounted on a custom 3d printed hat on its head and also a chest harness that has an accelerometer and a gyroscope. So basically just detecting motion. And they use QR scanners to detect each dog. So basically it'll scan in like this is Fido. Fido gets licked into this gate and then there's a timer of how long they're allowed to do this. And then they get rewarded with a treat. And I thought this is really cool. And they explained, so they really look at the dogs as collaborators. and i was wondering like why do you have an eeg and a harness with an accelerometer and all these like video detection not only obviously you want multiple kinds of data but they said each dog detects in a different way so one of them might sit down and that's where the gyroscope would come in and another might just like breathe really fastly so they're measuring um different kinds for different dogs and basically and they call it the sniff analysis sniff analytics sniff analysis um how they parse the data and line it up and it's actually really well done.

33:47I'm not going to go into the specifics, the sniff-cifics.

33:50Dr Niki: The sniff-cifics. Yeah, the specifics. But it's very cool how they, basically, they'll take all this data and they'll say, you know, if you get this specific set of combinations, that's a yes from this dog. And if the sample they train them on, you know, they know which ones have cancer initially as a training sample. That's how they figure that out. So like a lot of machine learning, it's speeding up and making more accurate what a human could do, right you could have a really good dog trainer understand the body language of the dog and figure out their behavior uh but this is probably trained from humans initially as well yeah yeah 100 i think from what i understand the most of the machine learning is coming from the behavioral analysis but it looks like it's also used to like sync up all the data and make outputs but it is it looks like it's very much just trained on humans and then like do it times Well, and that was the part that I wasn't sure about when, when I first sent this to you, I'm like, Dr.

34:46Dr Niki: Nicky's going to tell me like, you don't need the AI to do this. You could just look at the dog. And so it's interesting that you looked at her like, no, it looks like this is actually a valuable addition. And I have this from like experience because we, we, we talked about on the show, but I do video analysis for behavior analysis for my goats. And we have like 6 ,000 hours of video and there's only so many that my undergrads can count. So like I actually really want machine learning to do a good job at this because this human hours that we could be doing something else. So yeah, that's a great example of that.

35:14I emphasize with this a lot.

35:15Dr Niki: All right. So how do we know this works? We know that there's research that shows that the dogs can smell this and that there's actually compounds, but how do we know this particular system is actually working? So even more proof, if you weren't convinced, they have published their phase two trials in the Journal of Clinical Oncology, and they report a 90 percent sensitivity for early stage cancer, which is even more impressive because you would assume there's less, you know, volatile compounds in an early stage. And this was done on about over a thousand participants. And so since they've published that, they started phase three trials.

35:49So this is the largest, largest scale in April, enrolling over 10 ,000 people. and the company is now targeting a million tests a year with just 30 dogs. So those dogs are like pumping those tests out. And they have plans for a U.S. facility in 2027 or 2028 pending regulatory hurdles because you have a combination of animal ethics here and FDA-ish stuff where you don't know if diagnosis sits really in FDA or not. But they're confident.

36:17Dr Niki: Yeah, and I was reading that in India there's less of a burden for detection style diagnosis like this versus the United States where you do have to get approved for that. So another reason that they're rolling this out in India. And from what you can tell, the dogs are happy with this. Yeah. So they have a very big emphasis on like, these dogs are really well treated. They adopt them and then they train them. And I would treat them really well too. If they were pumping out a million diagnoses, I would assume that they're, at least from the pictures and like from the setup and from what they say on their website there.

36:55And even in the publication, they emphasize like we see the dogs as collaborators. They go willingly to do this training. We don't force them, you know, and you know how dogs are. If they have a job, usually they like doing it. So as far as I can tell, it looks actually really smart. And I'm really inspired by this and I might integrate it into my own work.

37:12Dr Niki: That's amazing that this might actually end up helping your, your actual scientific work. I have to put DTNS in the acknowledgements or something. You don't have to do that. I always do anyway. But, yeah, while this study gave me pause, it seems like it's a furway down the line, you know. So thank you for telling us the tale. Yeah, I'm glad we sniffed out the truth here. Indeed. Dr. Nikki, if people want to find more of what you do, where should they go? I am over at NicoleAckermans.com, same handle on Blue Sky. And if you want more fun science and tech, you can check out my new show with Wen Tui Dao in the DTNS Family podcast.

37:54It is technical tea. Our new episode is going to be about eponyms. So next month, come check that out.

38:01Dr Niki: Fantastic. Thanks, Dr. Nicky. Thank you. I like this. It's got cute dogs. It's going to help fight cancer. What's not to love? um you there have been studies and there have been you know dogs that sniff cancer out people who had no idea that they had it this has been happening for decades now so if you can use a little ai if you can use you know um robots to actually improve it that is a good thing this is this is where i where i see ai and robots being good for humanity to figure out things cancer is such a scourge you know you know on health for earth if you can do something to help people get better diagnoses earlier, I'm all for it.

38:42Dr Niki: And 100%. Folks, we end every episode of DTNS with some shared perspectives. Today, Jeff in Knoxville, Tennessee has added context to the story about Pringles using a digital twin to perfect its chip stacking. Yeah, a lot of people like that story, even though it was real quick. Jeff wrote, last week, you covered the recent story about the makers of Pringles creating a digital twin of their factory to optimize their production. I I felt compelled to point out that Pringles has a long history of using technology like this. The shape of Pringles was optimized on a Cray supercomputer, a Cray 1, if my memory serves me, to give them a more aerodynamic shape and prevent them from literally flying off the production line at high speeds.

39:25Dr Niki: Back when I worked for Cray, this was my favorite story. Whenever someone asked me what sorts of important world-changing problems are being solved by supercomputers, i'd imagine that you're hard pressed to find a more technology technologically forward snack food than pringles i like pringles this makes me proud to uh to like a stackable chip this is cool we used a cray supercomputer to figure out how to make pringles better can we legally call pringles chips i don't there's debates on that i don't i don't know i mean then you gotta start to get into the whole english thing about crisps and chips and you You go down a rabbit hole.

40:03This is pretty cool, though. I mean, we're using it to solve problems that we really have. These chips are different. They're light. They fly off the machine. We need to figure out how to make them stick. Yeah.

40:13Dr Niki: No, I love it. Well, folks, we love to know what you're thinking about. So if you've got insights into a story, please, please, please share it with us at feedback at dailytechnewsshow.com. Big thanks to Jeff for contributing to today's show. Thank you for being along for Daily Tech News Show. You're the folks who keep us in business. If you want to directly support the show and get no ads, become a patron. You can do it easy. Patreon.com slash DTNS.

40:41The DTNS family of podcasts. Helping each other understand. Diamond Club hopes you have enjoyed this program.

40:56Dr Niki: Hey guys, it's Drew Afualo from the comment section. Listen, it's time to stop settling for hair products that just don't work. With all the prestige brands at Sephora, upgrading your routine is easy. If your hair needs repair, they've got the K18 Leave-In Molecular Repair Hair Mask. If you're trying to level up your wash day, the Kerastase Elixir All-Time Hair Oil leaves your hair looking ridiculously shiny without feeling heavy. And if you're into that glossy, healthy hair vibe, the Gisu Honey Infused Hair Oil is a must-have. Also, Sephora makes it super convenient with free shipping and same-day delivery.

41:30Dr Niki: Shop hair at Sephora and give your hair the upgrade it's been waiting for. This episode is sponsored by Aerie. If you know me, you know that my philosophy is all about looking good and feeling good every single day. And that honestly comes down to the first layer. I'm talking about bras. If your bra is digging into your shoulder, or maybe it's riding up in the back, or you are dying to take it off at the end of the day, then that bra is not the right fit. And what I love about Aerie is that they have 59 different sizes in really soft fabrics and really good fits. Because at the end of the day, your bra should hold you in and not hold you back.

42:05Visit Aerie.com to find the right bra and your perfect fit.

42:20Thank you.