Introduction
Are you curious how artificial intelligence will change the wines you drink? What’s the difference between AI-based technology and existing automated machines? Will AI eventually replace most people working in the vineyard?
In this episode of the Unreserved Wine Talk podcast, I’m chatting with Dina Blikshteyn, a lawyer who specializes in how artificial intelligence and machine learning is changing the wine world.
Note: Our discussion is not intended to be a substitute for professional legal advice and is for informational purposes only.
You can find the wines we discussed here.
Join me on Instagram, Facebook and YouTube Live Video
Join the live-stream video of this conversation on Wednesday at 7 pm eastern on Instagram Live Video, Facebook Live Video or YouTube Live Video.
I’ll be jumping into the comments as we watch it together so that I can answer your questions in real-time.
I want to hear from you! What’s your opinion of what we’re discussing? What takeaways or tips do you love most from this chat? What questions do you have that we didn’t answer?
Want to know when we go live?
Add this to your calendar:
Highlights
- What was Dina’s first memory of drinking wine as a poor college student?
- What’s the difference between automation and artificial intelligence?
- How is AI improving grape growing and winemaking systems?
- What’s the difference between newer AI-enabled machines and existing automated farming machines?
- Which types of technology are used with AI in the wine industry?
- Is there a risk of AI replacing human intervention in vineyards?
- Who owns the data associated with machine learning, and what are the risks with data privacy?
- What are the legal implications around the fair use of data obtained from the internet for AI training?
Key Takeaways
- I was fascinated to learn how artificial intelligence will change the wines you drink. AI can determine whether the grapes are getting enough water in the growing stage and other factors to optimize ripeness and avoid disease. That, in turn, will make for better quality wine related to those factors.
- Dina’s clarification of the difference between AI-based technology and existing automated machines was helpful. AI is a subset of automation that involves training models on data. It eventually makes new decisions and outputs, whereas traditional automation sticks to the rules you set for it.
- I’m glad to hear that AI will not eventually replace most people working in the vineyard. Dina makes a great analogy that when the calculator was invented, we still needed to know how to do math. It’s the same thing with AI; it’s just a tool.
Start The Conversation: Click Below to Share These Wine Tips
AI is a subset of automation that involves training models on data. - Dina Blikshteyn Click to tweet
About Dina Blikshteyn
Dina Blikshteyn is a partner in the Intellectual Property Practice Group in the New York law office of Haynes Boone. Dina focuses on artificial intelligence and machine learning, cloud computing, cyber security, web applications, algorithms, multimedia and video streaming, among other technologies. She is also a co-chair of the artificial intelligence practice at the firm. Prior to becoming a lawyer, Dina developed high-frequency trading systems that traded financial instruments on domestic and international exchanges.
Resources
- Connect with Dina Blikshteyn
- Diary of a Book Launch: An Insider Peek from Idea to Publication
- Wine Witch on Fire Free Companion Guide for Book Clubs
- My Books:
- Unreserved Wine Talk | Episode 80: What’s the Rosé Lifestyle? Wine Writer Jill Barth on Why You’d Enjoy It
- My new class The 5 Wine & Food Pairing Mistakes That Can Ruin Your Dinner And How To Fix Them Forever
Tag Me on Social
Tag me on social media if you enjoyed the episode:
- @nataliemaclean and @natdecants on Facebook
- @nataliemaclean on Twitter
- @nataliemacleanwine on Instagram
- @nataliemaclean on LinkedIn
- Email Me at [email protected]
Thirsty for more?
- Sign up for my free online wine video class where I’ll walk you through The 5 Wine & Food Pairing Mistakes That Can Ruin Your Dinner (and how to fix them forever!)
- You’ll find my books here, including Unquenchable: A Tipsy Quest for the World’s Best Bargain Wines and Red, White and Drunk All Over: A Wine-Soaked Journey from Grape to Glass.
- The new audio edition of Red, White and Drunk All Over: A Wine-Soaked Journey from Grape to Glass is now available on Amazon.ca, Amazon.com and other country-specific Amazon sites; iTunes.ca, iTunes.com and other country-specific iTunes sites; Audible.ca and Audible.com.
Transcript
Dina Blikshteyn (00:00):
There’s a lot of research that’s going on right now by the company called Covata. And what this company is doing, it’s pairing up with vineyards and it’s sending robots through the vineyards that take pictures of the grapes. And based on those pictures, they’re trying to figure out whether the grapes on the vine, the trees are actually getting enough water. So here AI is determining whether the grapes themselves are getting enough water.
Natalie MacLean (00:30):
Far more sophisticated because different grapes probably have different levels of the ability to absorb moisture. So looking at the grape health and the leaf health is far more sophisticated. And actually going right to the grapes and the leaves I would think that’s far more accurate.
Do you have a thirst to learn about wine? Do you love stories about wonderfully obsessive people, hauntingly beautiful places and amusingly awkward social situations? Well, that’s the blend here on the Unreserved Wine Talk podcast. I’m your host, Natalie MacLean, and each week I share with you unfiltered conversations with celebrities in the wine world, as well as confessions from my own tipsy journey as I write my third book on this subject. I’m so glad you’re here. Now pass me that bottle please and let’s get started.
Welcome to episode 242. Are you curious about how artificial intelligence AI will change the wines you drink? What’s the difference between AI-based technology and existing automated machines? And will AI eventually replace most people working in the vineyard? In today’s episode, you’ll hear stories and tips that answer those questions in my chat with Dina Blikshteyn, a lawyer who specializes on how artificial intelligence and machine learning is changing the wine world.
(02:15):
Speaking of the wine world, as I got close to finishing my new memoir Wine Witch on Fire: Rising from the Ashes of Divorce, Defamation, and Drinking Too Much I found creative ways to delay handing in the final manuscript to my publisher. I wanted to keep editing it to be perfect. Triple checking every detail. Perfection is procrastination in disguise. Perfectionism and competitiveness coiled together like a cobra and a boa constrictor. The first bites you with envy, the second squeezes the joy of life out of you. Together, they’re the undisciplined pursuit of more. But I also think it’s the struggle that counts. Now, there’s a person behind the person who is all about perfection and winning. This quieter self sees me striving and says relax, sister. I’ve got you. Do you struggle with perfectionism or competitiveness? Let me know.
Here’s a review from Diane Bader in Calgary, Alberta. “When we name what’s happening to us and how we feel about it, we often feel a release. That’s why memoirs are so powerful. People feel seen and heard in someone else’s experiences. Someone has put it into words for them.” Actually, she’s quoting me there. So inside a quote, all right, now I’ll continue with her review end bracket.
“She says this is on page 259, which has a lot of great stuff. During my own coming of middle-aged story, I discovered Wine Witch on Fire in a wine store in the Okanagan Valley after moving from Toronto. Coincidence? I don’t believe in them. I can say that once I dug into this book, I couldn’t put it down much like a good glass of Semillon. This book is not only filled with Natalie’s story of a horrific year that completely changed her life, but lots of great tidbits about wines, wineries, and witches. I found myself taking notes.
“Whether you love wines or just looking for a good memoir to read, Natalie’s story deals with her entire world imploding after the end of her marriage, as well as an online attack from people who envy her success. I highly recommend this book, which has several great insights and is in an easy to read style. It also pairs well with a patio and something called light and bubbly, but that’s just my personal preference”.
Thank you, Diane. If you’ve read the book, I’d love to hear from you at [email protected]. If you haven’t got your copy yet and would like to support it and this podcast, please order it from any online retailer no matter where you live. Every little bit helps spread this message. I’ll put a link in the show notes to all retailers worldwide at NatalieMaclean.com/242. Okay. on with the show.
Natalie MacLean(05:10):
Dina Blikshteyn is a partner in the intellectual property practice group in the New York Law office of Haynes Boone. I should note that although Dina is a lawyer, our discussion today is not intended as legal advice and is for informational purposes only. Dina focuses on artificial intelligence and machine learning, cloud computing, cybersecurity, web applications, algorithms, multimedia and video streaming among a host of other technologies. She’s also the co-chair of the artificial intelligence practice at the firm. Prior to becoming a lawyer, Dina developed high frequency trading systems that traded financial instruments on domestic and international exchanges. She joins us now from New York. Dina, so great to have you here with us. Welcome.
Dina Blikshteyn (06:01):
Thank you, Natalie, and thank you for that wonderful introduction. It’s my pleasure to be with you as well.
Natalie MacLean (06:08):
Awesome. Well, you just have a wonderful background. I’m fascinated by this topic. But before we dive in, I want to hear about some of your personal wine moments because you are a wine person as well as being an expert on AI and law. So tell me a little bit about when you were in college, perhaps budget strapped and trying to open a bottle of wine.
Dina Blikshteyn (06:29):
Yes, that’s one of my first memories of drinking wine and being a poor student as most of us are or were. We were taking a road trip. It was a group of four of us and of course we picked up some wines. And when we got back to the hotel, we realized we didn’t have a wine opener. Of course, it’s being 11 o’clock at night and we were trying to relax after a long day of sightseeing. I believe we were in Canada at the time. We were just trying to figure out how to open this bottle of wine. So me being an engineer at heart with an engineering degree, of course I came up with the solution of using a pen and a boot. So it was very creative. We actually decided to push the cork inside the bottle of wine, which is something you’re not supposed to do, but we didn’t really have that many options at that time.
Natalie MacLean (07:26):
Desperate times required desperate measures.
Dina Blikshteyn (07:30):
And in a long story short, it did work.
Natalie MacLean (07:32):
It did. So you said a pen and a boot.
Dina Blikshteyn (07:35):
And a boot.
Natalie MacLean (07:35):
Okay. So you’re using the boot as a hammer kind of thing.
Dina Blikshteyn (07:38):
The boot was a hammer to push the cork right inside the bottle.
Natalie MacLean (07:44):
Very good ingenuity right from the beginning. All right. You’ve also mentioned that in various pictures that have been taken, family shots throughout the years, what’s been unusual or what does your family say about these photos?
Dina Blikshteyn (07:58):
Well, this is my older daughter who always teases me about this. It seems like every time somebody takes a picture of me in a social setting, I always have a bottle of wine next to me or a glass of wine. And a couple of weeks ago, we had a conference with other female attorneys in AI actually and we were all in the restaurant trying to get to know each other. And we asked for a waiter to take a picture of us. So this was after dinner. The table was empty. Of course, there is four glasses of water and there’s a glass of wine next to me.
Natalie MacLean (08:35):
Of course.
Dina Blikshteyn (08:35):
Of course.
Natalie MacLean (08:37):
Well, at least you’re consistent.
Dina Blikshteyn (08:39):
That’s very true. Actually, my older one also drew a picture of wine glass that’s in my office right now.
Natalie MacLean (08:47):
Oh, wow. So cute. And is she making a 3D wine glass?
Dina Blikshteyn (08:52):
Yes. So apparently there is new toy on the market – maybe not so new; I found out about it last week – w here it’s a 3D pen. It’s sort of like a glue stick. So you can pick colours that go into it and it heats it up, and then you can create a 3D object.
Natalie MacLean
Oh, wow.
Dina Blikshteyn
So one behind me, this is the one she created yesterday.
Natalie MacLean (09:17):
Oh, wow.
Dina Blikshteyn (09:18):
There we go. It’s a piece of art.
Natalie MacLean (09:20):
And if you’re listening to this on the podcast, you can watch the video version to see the picture.
Dina Blikshteyn (09:24):
Yes, it’s very cute. But once they found out I was doing this podcast, so my younger daughter is actually creating a glass of wine in the…
Natalie MacLean (09:35):
Well, they’re very supportive. That’s good. Alright, I love those stories. Okay, Dina, let’s dive into our topic. I’m going to take a stab at clarifying the difference between automation and artificial intelligence, and then you can correct my errors. As I understand it, automation focuses on repetitive tasks and rules and creates systems or software for computers or robots to follow them. This is my Wikipedia research showing through here. Artificial intelligence or AI is focused on non-repetitive tasks and getting computers or robots to make decisions on their own based on human input and machine learning. So in a sense, AI keeps growing and changing on its own whereas automation does not without additional human intervention. How would you correct or expand on that?
Dina Blikshteyn (10:21):
Okay, so my take on AI. AI is a subset of automation because once you start looking at AI, AI is being trained on data. So the more complete the data is, the better AI would be trained. So what you’re doing essentially is having an AI model and you’re passing data through it over and over and over and over and over again until you’re getting a result that is right. Once that model is trained, it goes into the real world and then you have data going in and they’ll determine the result for you. So it’s automated but it’s not necessarily based on a set of rules. It’s more based on how during the training stage the model has been taught to think based on this initial data set.
Natalie MacLean (11:13):
Fascinating. Okay, so let’s dive into how AI is related or involved with wine these days, including I mean all aspect: grape growing, wine making, distribution, recommendations, and even food pairing. So let’s focus on the vineyard first. So we’ve heard about soil sensors calibrating irrigation in the past. How is this different when AI machines take over or how is this improved upon with AI?
Dina Blikshteyn (11:40):
Okay. Well one, there’s a lot of research that’s going on right now by the company called Covata. And what this company is doing, it’s pairing up with vineyards in Napa Valley and it’s sending robots through the vineyards that take pictures of the grapes. And based on those pictures, they’re trying to figure out whether the grapes or the vine, the trees are actually getting enough water. So in your example, you determine whether the soil is getting enough water. Here, AI is determining whether the grapes themselves are getting enough water. So I guess it’s how you’re looking at it, Natalie. Do you want to know if the soil is wet enough for the roots or if the water actually goes through the grapevine and reaches the grapes?
Natalie MacLean (12:29):
Right, far more sophisticated because different grapes probably have different levels of osmosis or ability to absorb moisture. So looking at the grape health and the leaf health is far more sophisticated and accurate I would think because we are hoping wet soil leads to vines that are healthy. But if you’re actually going right to the grapes and the leaves, I would think that’s far more accurate.
Dina Blikshteyn (12:55):
And I think that actually the two processes are complimentary, right. So you need moisture level in the soil to have the grapes grow. But once you have the grapes, you can actually figure out if the grapes are getting enough water or if the water is going towards the leaves. So the two processes are complimentary and you may actually need both at the vineyard.
Natalie MacLean (13:17):
Sure, absolutely. Especially in the early stages when you don’t have fully fledged grapes and leaf canopy. Now we’ve also heard of drones mapping vineyards and heat resonance maps and so on. How does AI change and improve that process?
Dina Blikshteyn (13:34):
Well, one way that I can think of. You have the drones that do the mapping and then you plant the vineyards. But how do you know whether those vineyards will be successful later on? So what AI can do, it essentially can analyze different drone maps and whether the vineyards have been successful further down the road. And once you have that particular model trained, it can do predictions. What types of places would be where the vineyards is more or less successful?
Natalie MacLean (14:07):
So more learning. And then back to the grape leaf, the pictures that it’s taking. This is an autonomous machine that moves through the vineyard independently that’s based on photo analysis whereas is the mapping of the vineyards based on predictive data? Like this type of soil with this soil composition usually yields the best vines for Cabernet versus Chardonnay?
Dina Blikshteyn (14:31):
Potentially, right. It’s all about what you program the model and the machine to do. So a lot of it is just data analytics based on past examples. And if you want to do which soil is better for Cabernet versus the Chardonnay or any other type of wine, you analyze the previous data, train the model, and then it will predict the likely result for you.
Natalie MacLean (14:58):
Do these machines operate autonomous continuously throughout the vineyard all year round? Are they always sort of taking different measurements? How do they work?
Dina Blikshteyn (15:07):
It’s really up to the winery. So the technology of just autonomous machines have existed for a long time. So if you look at John Deere for example, they have autonomous machines going all year around, but it’s more farm related. So here the novelty is at least with Kubota, who also makes similar machines in Japan, is applying them to wineries and more take to taking the image of the grapes rather than planting and harvesting.
Natalie MacLean (15:38):
And how is that different? So they’re analyzing grape leaves, but how is that different from analyzing cabbage leaves? I mean, I know they’re two different plants but why is Kubota focusing on the wine industry and how is the technology really different from what maybe John Deere is doing with cabbage?
Dina Blikshteyn (15:57):
So I don’t know exactly what John Deere is doing, whether they are using AI or if it’s more of an automation rule based machines that we discussed earlier? But there’s several types of technologies here. One is just autonomous driving and that point you’re teaching machines how to drive autonomously, whether it’s through the vineyard or through the farms or if you want to drive on the street like Tesla, for example, or other companies that are complete autonomous driving on the streets. So that’s one component. The other component is actually directing it at a wine industry. And Kubota, for example, is particularly interested in how to improve the wine industry. So maybe it’s an untapped market or maybe they have AI technology that they think is more specific to wines. We don’t know. But there’s a lot of research going on in that space.
Natalie MacLean (16:52):
And I would think that the wine market would be a good one because the margins on wine as a finished agricultural product would be, I would think, just much higher than apples or any unprocessed fruit or vegetable. There’s just so much value add that goes into wine that I would think that the inputs, the computers and technology that you could use to produce the wine, including AI, there’d be more margin to invest in that kind of equipment.
Dina Blikshteyn (17:18):
That’s very possible.
Natalie MacLean (17:19):
Very possible. Okay.
Dina Blikshteyn (17:21):
To some extent, it’s all about how to make things better, faster, and more profitable.
Natalie MacLean (17:27):
Sure. And how large are these machines? I’m imagining Terminator kind of walking through the vineyards and it’s red eyes taking shots of the leaves, but what do they look like and how big are they?
Dina Blikshteyn (17:39):
The machines come in different sizes. Again, I’m just going back to the John Deere references that I think more people can resonate with that, right. There are big machines that are used at the farm that are just ginormous, and then there is smaller machines. And I think for the vineyards, when you go between rows of vines, there’s not that much space. So you are going to need a machine that can essentially go between the wines and then take pictures. So we’re not looking at those ginormous expensive machines. It’s probably something that’s small.
Natalie MacLean (18:13):
Yeah, I would hope so because there’d be a risk of compacting the soil, which we know is not good for microbial health and vineyard biodiversity. But I suppose I guess they’re taking that into consideration.
Dina Blikshteyn (18:26):
All right. I would hope so as well. Yeah, we don’t want to have AI linked to red wine, right.
Natalie MacLean (18:32):
No. That would kind of defeat the purpose. Water these but then they get smooshed. I know different machines have different costs, but what might be an average cost or a starter cost of one of these Kuboto machines? Are we talking like $5,000, $50,000, $500,000?
Dina Blikshteyn (18:50):
That’s a question that I can’t answer. I just don’t know how much those machines will cost.
Natalie MacLean (18:55):
Okay. I was just curious. And do you have any sense of how many wineries are using them or what the adoption rate is, or are they just really test cases right now?
Dina Blikshteyn (19:06):
As far as I understand, right now we are in the beginning of the research phase. So we’re in the testing phase to figure out how those machines would be used and how they’re being used. I haven’t seen wide adoption yet. However, as both the automation technology is improving with autonomous driving as well as image technology with the machines taking pictures and analyzing them, I suspect the adoption will occur in the future by many wineries.
Natalie MacLean (19:36):
Probably like self-driving and electric cars. As the cost comes down and the technology becomes more widespread and more accessible, there’ll be higher adoption rates. Have you heard of any success stories from any wineries using the technology? They experienced a 30% cut in costs or any sort of early results?
Dina Blikshteyn (19:56):
I have not just because right now it’s still in the testing phase. It’s too early to test. But I mean, if you look at the way AI technology is going, especially with Chat GPT, there is a huge adoption and people are testing it out for different use cases. So I think once the technology is successful, it’ll be like a switch. The same thing with a Chat GPT. It was slow, slow, slow. Then Chat GPT came out and now everyone is using it. Then it’s sort of also threw a lot of companies and industries into a flux. So I anticipate if these machines are successful, we’ll see the same sort of adoption here, particularly if the wineries will be increasing the profit margins due to this technology.
Natalie MacLean (20:41):
Absolutely, yeah. Do you think there’s a risk that these machines will maybe not completely replace all humans in the vineyard, but for the most part replace all human intervention in the vineyards?
Dina Blikshteyn (20:53):
I don’t think so. It will definitely make some things easier and faster, but I’ll use the same example that I tell my kids and my clients. When you calculator was invented, you still needed to know how to do math. It’s the same thing. AI is just a tool. It’s very good at analyzing data. It can automate some tasks, but it’s only as good as how it’s being programmed. And for that you need human intervention. So replace some jobs probably. Will it shift to people performing other jobs? Definitely.
Natalie MacLean (21:31):
Yeah. I’m a bit of a tech geek as well and I listen to tech podcasts and the advice I’ve heard so far to young graduates is, you know, won’t be replaced by AI but you could be replaced by someone who knows how to use AI to augment what they do on their job. So it’s a good thing to know how to use the tools.
Dina Blikshteyn (21:50):
Oh, absolutely. It’s a great tool to have. And once you know how to use it, it can make your job faster. And I also give you a competitive advantage.
Natalie MacLean (22:01):
Yes. Cool. All right. So now we’ve talked about how AI depends on human based input, especially with your legal background. I’m just curious your opinion only for information purposes, but who owns that input? Is it the winemaker who purchased the machine or software, the company who made it, or both? Especially when we’re dealing with proprietary blends and vineyard grapes and all the rest of it. What kind of discussion has there been on that and are there any laws or precedents yet?
Dina Blikshteyn (22:31):
It’s actually a very hot topic because with a lot of these AI models. The models themselves are not new. It’s all about the data. So you need the data for training. So essentially the more data you have, the more accurate the model will be. So it’s becoming a very hot topic. Who owns the data and who can use the data.
Natalie MacLean (22:53):
Is there anything emerging or it’s just still completely unclear? When people use Kubota, do they own their data – the winemakers – or is it all going back to Kubota?
Dina Blikshteyn (23:03):
Right. So the way these machines typically work, a lot of these machines come pre-trained in which case Kubota will get an AI model right on the machine. They are already pre-trained on third party data. And it’s likely whoever owns that machine owns that data. But what a lot of people don’t realize, those machines can also be what’s called fine tuned. So what you have here, you can have your own data that you used to train that last layer of the machine to make the machine behave in the way you want. So that data belongs to you.
Natalie MacLean (23:42):
To the winemaker?
Dina Blikshteyn (23:43):
The winemaker.
Natalie MacLean (23:43):
The winemaker who’s inputting it.
Dina Blikshteyn (23:45):
And then the other question becomes, well you’re using this AI machine, are you buying the machine or are you buying access to the machine that’s also being used by other wineries and other people? And that is a business decision. So it’s similar to Chat GPT. If you want to go log in, use the Chat GPT model along with anyone and everyone else, or do you as part of the company want to buy your own version? And then that version and that data stays with you.
Natalie MacLean (24:18):
Right. So people have to kind of know what they’re buying and what their contracts say when it comes to that. Do you see any parallels with what’s happening now with this data and input with some larger agricultural companies selling proprietary or genetically modified seeds to farmers and then the farmers become dependent on the seeds and then they can’t grow any other crops because these seeds are – whatever it is – trademarked or copyrighted or whatever they are legally. Do you see any risk of that for winemakers becoming so dependent on these AI machines that then they really are entrenched with these AI companies?
Dina Blikshteyn (24:57):
It’s a potential risk. It’s a risk because if AI makes your life easier, all of a sudden somebody else can controls your revenue. And again, a lot of it is just a business decision. Do you want AI to supplement your entire process for growing the ones? Or are you just using it as a data prediction tool? And a lot of that is really contractual. We actually see people get in trouble and this is going to be a reoccurring theme. You need data to train AI.
Natalie MacLean (25:27):
Yes.
Dina Blikshteyn (25:28):
So you can have third party data, you can have your own data, but then you don’t have enough data to get a complete picture. You just need more use cases, right.
Natalie MacLean
Yes.
Dina Blikshteyn
And then people start going on the web trying to get the data, and they may not have permissions to use that particular data. So the problem with all these AI models is once the data goes into the machine, you can’t take it out because it’s used to train different weights inside the machine. So now you have this machine trained on data that you can’t really use. And if there is a court order forbidding you from using that data, you may just need to start from scratch.
Natalie MacLean (26:09):
Ooh, wow, that’s drastic.
Dina Blikshteyn (26:10):
Yeah, that’s a lot of money going down the drain. You have to retrain your own model without the data. But that’s where people can really get in trouble, sort of get this data they can’t use that they need for their business, and they put into the model without permissions.
Natalie MacLean (26:27):
And, either from your perspective of what you’ve heard discussed in the legal community, is it fair use to train a computer on data that’s widely available on the internet? Because as I understand it, the AI machines are not reproducing that data. It’s not like they’re sort of plagiarizing and saying, here is the data. They’re just using it to learn. So is that fair use or fair dealing?
Dina Blikshteyn (26:51):
That’s actually a hot topic right now. Snd if anyone wants to Google, there’s a Getty case that’s going on and that’s the argument they’re making. Well if I’m using data to train AI, is that fair use? There’s no court decisions yet. It’s something that we’re all looking at and see which way the law will fall. Because depending on which way the courts come out, and I’m sure whatever the decision will be made, it will be percolated to up the line to the Federal Circuit, if not higher. And then just trying to figure out whether training AI with somebody else’s data is fair use or not, right. Now, at the same time, this is not a problem that exists in Europe, right. Because there’s a counterpart to the Getty case that’s going on in England right now and they don’t have a fair use defence. It may also be something that’s really just country specific.
Natalie MacLean (27:44):
So England doesn’t have a fair use law or provision the way the US does and Canada has fair dealings. So the UK doesn’t have that provision?
Dina Blikshteyn (27:53):
As far as I understand.
Natalie MacLean
Okay. Interesting.
Dina Blikshteyn
But they don’t have it at this point. It’s just too early to say. Rhere’s definitely litigation going on. It’s just we don’t know what the outcome will be.
Natalie MacLean (28:04):
Interesting. Do you have an opinion one way or the other in terms of if AI is learning off the what’s on the internet, is it fair use or is it not?
Dina Blikshteyn (28:14):
It’s a loaded question because just because something is on internet, it doesn’t mean it’s not copyrighted or protected and that it’s available for use.
Natalie MacLean (28:23):
Yes.
Dina Blikshteyn (28:24):
I mean in my opinion as an attorney, if you are using somebody else’s data, make sure you can either use it or you compensate the person for it.
Natalie MacLean (28:35):
Even if you’re not reproducing it?
Dina Blikshteyn (28:36):
Right. Does that always happen? No. Because I mean I also think as attorneys we’re more aware of what can and cannot be used, but if you’re telling an engineer don’t use certain type of data and that’s the data they need to succeed that’s a completely different calculus.
Natalie MacLean (28:54):
Okay. Wow.
(29:00):
Well, there you have it. I hope you enjoyed my chat with Dina. Here are my takeaways. Number one, I was fascinated to learn how artificial intelligence will change the wines we drink. AI can determine whether the grapes are getting enough water in the growing stage and other stages as well as other factors to optimize ripeness and avoid disease that in turn will make for better quality wine related to those factors. Dina’s clarification of the difference between AI-based technology and existing automated machines was helpful. AI is a subset of automation that involves training models on data. It eventually makes new decisions and outputs, whereas traditional automation sticks to the rules you set for it. And finally, I’m glad to hear that AI will not eventually replace all people or most of them working in the vineyard. Dina makes a great analogy that when the calculator was invented, we still needed to know how to do math. It’s the same thing with AI. It’s just a tool to help us do our jobs better.
In the show notes, you’ll find the full transcript of my conversation with Dina, links to her website, the video versions of these conversations on Facebook and YouTube live, and where you can order my book online now no matter where you live. That’s all in the show notes at NatalieMacLean.com/ 242. Email me if you have a sip, tip, question or you’ve read Wine Witch on Fire at NatalieMacLean.com.
If you missed episode 80, go back and take a listen. I chat about the particular allure of wines from Provence, especially Rosé with Jill Barth. I’ll share a short clip with you now to whet your appetite.
Jill Barth (30:44):
There is a preference for a pale Rosé. I think that that’s the consumer’s idea that it embodies freshness and lightness. There also seems to be misconception that it’s going to be sweeter if it’s darker and that’s not true at all. I’ve heard people I’m sharing wine with and they’ll see a dark Rosé and they’ll think oh I don’t like sweet wines. But it’s to do with the varieties of the grape and how long that juice has any amount of skin contact.
Natalie MacLean (31:10):
That makes total sense. Yeah.
Jill Barth (31:12):
People do seem to like the light ones these days for I think, reasons of aesthetics, not necessarily that it influences the flavor as much as you might think.
Natalie MacLean (31:19):
Sure. And is there anything to the fact that if it’s a darker rose, it’s going to be more full body? Did it get more skin contact, therefore more flavour? Or is that too a generalization that doesn’t always play out?
Jill Barth (31:33):
It probably doesn’t always play out, but it would be true that darker skinned grapes that experience more skin contact during the wine making are going to impart more of that colour.
Natalie MacLean (31:48):
If you like this episode, please email or tell one friend about it this week, especially someone who’d be interested in the wines, tips, and stories we shared. You won’t want to miss next week when we continue our chat with Dina. Thank you for taking the time to join me here. I hope something great is in your glass this week, perhaps a wine recommended to you by a human rather than a bot.
You don’t want to miss one juicy episode of this podcast, especially the secret full body bonus episodes that I don’t announce on social media. So subscribe for free now at NatalieMacLean.com/subscribe. Meet me here next week. Cheers.
