Instant Genius - Can AI ever really be ethical?

Episode Date: October 8, 2023

Artificial intelligence has grown drastically in recent years, entering everything from art to medicine. We talk to Nigel Cannings, an expert in artificial intelligence to better understand the ethics behind AI and copyright, working rights and energy consumption. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Ambition comes in all shapes and sizes. At First Citizens Bank, we roll with your goals because we're built for what you're building. Fit for your ambition for Citizens Bank. Study and play. Come together on a Windows 11 PC. And for a limited time, college students get the best of both worlds.

Get the Unreal College deal, everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 3.3. 365 premium and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller. Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m.m.S. slash college PC. It's peak pollination season, and my business is scaling fast. To keep the nectar flowing, I need a phone plan with top priority data speed.

That's why I chose GoogleFi wireless. My connections stay strong even when the hive is buzzing. Plus, unlimited plans started $35. a month. Now that's a deal that doesn't stay. Explore GoogleFi wireless plans today. Plus taxes and government fees. GoogleFi wireless is not subject to data traffic deprioritization during times of high network usage. This podcast is sponsored by name, audio and focal. Streaming has made music more accessible than ever, but true listening is about more than ease. It's about quality. British audio experts name audio, alongside French acoustic specialist focal, combine handcrafted tradition with

cutting-edge innovation and high-end materials, delivering digital precision with analog warmth. So you can experience exceptional sound at home. Music just as the artist intended. Visit nameadio.com to learn more. From BBC Science Focus magazine, this is Instant Genius, a bite-sized masterclass in podcast form. I'm Alex Hughes. staff writer at BBC Science Focus magazine. This week, we're talking about the complicated world of ethics in artificial intelligence. As a technology gets more advanced, a conversation is emerging. How do we police AI? Is it breaking copyright law?

And can it ever truly be a supportive worker? I'm joined by Nigel Canning's to discuss this topic. He's an expert in the ethics of AI and CTO of intelligent voice. He explains how he can address the ethical issues of artificial intelligence and how best to police it going forward. Artificial intelligence is, it's seemingly everywhere now, it's being applied to a huge range of industries. Does that raise a genuine fear about future job markets for humans or is it not actually as much of a problem as we like to make it out to be? Well, I think we've seen this with technologies before. I mean, I can remember going back when word processing first came out.

and there was a huge outcry about the fact that millions of typists around the world were going to lose their jobs, you know, typing pools and sectors and so on. I think, you know, any new technology is going to have some sort of shift in the workforce, and there's always the claims of revolution every time that comes round. So I think we're treading a familiar path here. I mean, inevitably, the job market will change. I mean, any new technology dictates that, really. I think in this particular area, though, there's a couple of interesting things that come out.

In many industries, what we're going to see is AI is going to help existing workers rather than replace existing workers. So people who work in the creative arts, people who work in copywriting, people who work in legal and accountancy, they're going to see their productivity greatly improved by this. I think initially companies are going to try and replace workers with this stuff, but we're already seeing that when you try and generate news articles from AI using large language models, they generate far too many hallucinations. It looks plausible, but it's actually rubbish.

So already people are retreating from that. So I think we're definitely going to see enhancements there. I suppose my concern is that too many employers are kind of thinking replacement immediately. And what that's going to do, if we're not careful, is leave us with a dearth of experts. And this whole idea of kind of expertise is one that's weighing on my mind a lot at the moment. Because if you look at say call center scenario, we've been replacing call center agents for quite some time now,

even with earlier generations of AI on how do you handle a question that's coming in from a customer, you know, how do I reset my password, whatever. But call center agents use that as training. So when you get to the really complicated questions, you've got a call center agent that over the years has learned how to deal with customers and can eventually reach a level of expertise. Same could be said for lawyers and accountants. If we cut out that early training of humans in a call center environment or in a legal environment or an accountancy environment, we run the risk of finding ourselves of actually losing an entire generation of experts. the same way we've seen COBOL programmers suddenly becoming really important because banks have realized there's no one to maintain their mainframes.

So, yes, there will be a change in the job market. I think it should be iterative rather than revolutionary. But I think those companies that do try and kind of drop human workers immediately in replacing with AI are actually going to find over time that they've bled the expertise out of their organizations. It's the ones who use it iteratively rather than in a revolution. way, I think, who are the ones who are really going to benefit from it. I mean, you touch on it a little bit there. I think it's quite interesting that a lot of

the work that's been done through AI on paper can look amazing or from a glance it can look amazing. And then you look at the finer details and you start to see the flaws or the hallucinations in it. And I wonder if once businesses start to replace human workers with AI, if it's then as you go through the time and you start to notice all these inconsistencies or you get to these difficult challenges you wouldn't have initially considered before, that's when it starts to fall apart a little bit. Well, I certainly hope it does, but I suppose it requires people who understand the technology to try and explain these problems up front rather than before people start to discover it themselves. I mean, there's a quite a well-known case of a New York law

firm who ended up generating a brief that was put in front of the court, which was generated using chat GPT and leaving aside the security issues and the confidentiality issues and so on, it generated a whole lot of case references, which were just wrong. I mean, they looked incredibly plausible and they were wrong. And the court sanctioned the lawyers for it and find them for it. And you would have thought that that would be enough to stop people rushing into it straight away, but they're not. So I think that, yes, a lot of companies are probably going to come unstuck.

I think there's going to be a lot of lawsuits that come out of it. We're already seeing suggestions of people being libeled by AI. So after this initial peak of excitement, I think it's only those companies who actually do understand the risks are the ones who are going to prosper from it. It raises, I think, quite an interesting topic there. I want to touch on copyright in a little bit, but there is also this issue, I think, with AI around who's responsible.

So, you know, with this law case, for example, you can't just say, well, this AI mess up and you can't make an AI system sign an NDA. You can't make it do any of these kind of things. It's not culpable for its mistakes. So who does that all fall on when things do go wrong? And yeah, and this is the thing. I don't think anyone's really sat back yet and thought, well, if I start answering questions using this stuff and it makes a mistake, who do I blame for it?

if most of the models which are out there at the moment, be the open source or closed source, have got so many caveats around them in terms of their usage, it's going to be quite difficult, I think, for a company who's relied on an open AI model to go back to Open AI and say, well, your model libeled one of my customers, they'll say, well, we warned you about this. We told you about hallucinates. So ultimately, the people are going to end up being responsible are probably the people with the biggest pockets. honestly. So I think there are going to be, there are going to be quite a lot of lawsuits against people like OpenAI, but also those companies who are using this stuff indiscriminately,

at the end of the day, they've got legal obligations within the European context. They've got very significant obligations now around the New Digital Services Act. We've got GDPR. You know, you cannot use a machine to make decisions about people unless you're, can explain how that decision was made. And one of the huge issues, I think, at the moment, is we're not building explainability into these models. The model can't tell you why it reached the decision it reached. And actually, that's just, you know, prima facie, a breach of European law if you're using this stuff to start making decisions on people. So, yeah, a lot of companies are suddenly going to find themselves tripping over very quickly. But unfortunately, if they're

using this stuff at scale, they may have made a lot of mistakes with a lot of customers before they realize that those mistakes have been made. So I think some of the fines we see coming out of using this stuff could be pretty eye-watering. And we're talking about the, I guess you described it as the end result of AI, but I think it's quite interesting to look at the training side of it as well, because, I mean, most of the AI models that are well-known, the chat GVTs, the dali's, all of these kind of ones, they're training. on information that was gained from the internet with all of its, I guess, its biases, some misinformation's in there and some prejudices,

these different kind of things that you just gain when you grab a load of information from such a wide variety of places. And then that then can push out on the other end things that are wrong or that maybe one AI might not see as wrong, but a human might see as wrong. So that is going to lead to even more issues on that side. Yeah, I mean, it's really hard to build a model which is unbiased. And the reason I say that is because none of us can actually agree on what bias means. I mean, it's humans we can't agree it.

We argue about correct political stances. We argue about which terms should be used or not be used. We argue over pretty much everything. So if you have a machine which is just spanishable, out text, which is effectively what these things are doing. I mean, I know we're looking at a wider context around generative AI with potentially audio and video and images as well. But yeah, it's difficult to get the machine to understand how to be human.

And so I think what that means is we have to be quite prescriptive in terms of the type of things that we ask this stuff to do. there are ways of, you know, you can use it for pretty narrow tasks. You can say, you know, what's the sentiment of this particular piece of text? That works pretty well. But if you say, you know, write me an essay about the rights and wrongs of X, Y and Z, you're going to start generating results which are definitely off the charts. So until we can find a way of training models in a way that people generically agree is unbiased,

yeah, you're going to have the horrible bits of the internet being spewed out to people who, frankly, are just not ready to receive it. They don't often understand what this stuff has been trained on. So it's not a surprise that some of what we get out is pretty horrifying. The other, I guess, issue that this form of training brings up is the ideas around copyright in, I guess, two issues that who owns the images that, if, I mean, images, texts, audio, whatever it is that you're creating, who owns the output? And should the models be training on images that they don't own the copyright to in the first place? What is the sort of best course of action when we start to talk about copyright in these kind of places? Well, one of my favorite recent developments is the fact that, sorry, OpenAI have released something called the GPT bot. So when, obviously, in order to gather training data, what the big companies have been doing is hovering in the internet in various.

various ways and they're scraping data from all over the internet. And of course, now they've got that data. So the Googles and the Open AIs of the world have already done their hoovering. Of course, Google doing it for a living. So what Open AI have done is released this thing called GPT bot, which says, I am a bot from OpenAI and I would like to scrape your data, please. Is it all right if I do that? And it, you know, kind of classic case of, of bolting the stable door here. We've already got all your data. So, effectively, they're trying to put in place something which stops other people getting

hold of it, which presents a bit of an interesting ethical dilemma, really, that should we really be in a world where we've allowed a few big companies to scrape the internet and then allow those big companies to turn around and say, well, we don't think anyone else should be allowed to do it. You know, seems to me to be a bit of a, you know, a surprising. place to find ourselves in. But in terms of the rights and wrongs of the data that's been scraped in the first place, the big companies anyway relying on a doctrine of fair use. And it's something which is more predominant in US law than in Europe. But there is still a similar sort

of right within European copyright law, effectively saying, well, you can use other people's information as long as you're not using it literally, if you're using it to create derivative works, it's okay to use that. That is considered fair use of somebody's work. And what we're seeing is that the initial cases on the generation of data from their staff is leaning towards the idea that actually the people whose work has been scraped do not have rights in this case. Because the images that are being produced by generative AI, let's use images as an example, those images are not actually copying the original.

And that's the point about copyright. Copyright is about reproducing an original image. So if, for example, somebody reproduced a work by Tracy M in, literally, that would be a breach of copyright. But having a piece of generative AI produce a work which is in the style of Tracy Emin is not. And actually, if you think about it,

if a human produced a pastiche of something, they produced something which looked like a Tracy Eminen work, as long as they didn't pass it off as Tracy Emin, then we wouldn't expect that to be a breach of copyright. So why are we suddenly saying that actually, just because the machines do it, that is automatically a breach of copyright. So I think a lot of these ideas have been around

in the human realm for a long time, the difference is what we've done is we've empowered millions of people to do this. I can't pick up a paintbrush. I wouldn't know where to start, but I could actually now produce something which looked realistically like a Tracy M in work using generative AI. The underlying concepts haven't changed. So I'm not sure really that it's as bigger deal as people are making out. I mean, I know that everyone's terribly upset about it. But actually, it's not as big a deal. The more interesting question, I think, is who owns the copyright in the thing that's been generated? That's the more interesting side of it. And there's actually recently even a case in the US. The US has got a kind of peculiar idea of how copyright works. So anywhere in the

world, you create something and copyright automatically subsists in it. So, you know, I write a book, it's copyright. I don't have to do anything with that. But in the US, you've got the idea that you can register that work. And it used to be you had to do it. I think I'm until about in 1976. If you wanted something to be copyright in the US, you had to submit it to be effectively,

you get a tick and a certificate and saying that this is copyright. So somebody decided to test this idea of whether AI art could be copyright. guy called Stephen Thaler, and he produced a work of art, let's call it that, submitted it to the US Copyright Office and said, this was generated autonomously by a computer, and I wish to register it as copyright in my name. Now, they rejected it. I don't think that was a great surprise. And there was a court case about it. He sued the US Copyright Office. and the case was rejected on technicality. And the technicality was that he said it was generated autonomously by a machine.

Now, what the judge was suggesting was, is that if there is an element of human input to it, and if he'd said that in his application, that actually there may be a case for saying that someone who has generated something maybe just using a really clever prompt could own the copyright in that. And I think this is where the next level is going to be coming in here, which is how much human interaction do you need with the computer to say that something is actually a human-owned work of art

rather than something which is just owned by no one. And I think we're sitting in a bit of a kind of legal no-man's land at the moment. So I think something generated autonomously by a computer, even probably just using a simple prompt, nobody owns. But when people start to incorporate AI into the creative process, that's when we're going to see humans saying, well, actually, that's all mine and nobody else is. You said this place was steps from the water. We just haven't found the steps yet. How much did we save? Enough.

Enough to get lost. Or you could book a stay with Hilton. Welcome to your oceanfront room. Just steps from the water. The Hilton sale is on now. Book on Hilton.com or the Hilton app and save up to 20% to get the stay you expected. When you want savings, not surprises.

It matters where you stay. Hilton for the stay. When you need to build up your team to handle the growing chaos at work, use Indeed-sponsor jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications, and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75-sponsored job credit at Indeed.com slash podcast.

That's Indeed.com slash podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed sponsored jobs. This podcast is sponsored by Name, Audio, and Trogel. With over 100 years of combined expertise, name and friends. name and focal have been bringing music to listeners just as the artist intended. Since day one, this mantra has shaped every innovation in hi-fi design, technology and acoustic engineering, balancing craftsmanship and tradition with pioneering thinking. Name audio pushes cutting-edge technology to ensure digital precision whilst sustaining Pratt,

pace, rhythm and timing, the elusive quality that makes music feel alive and gives it a emotional texture. Today, in partnership with French acoustic specialist focal, name audio creates systems that deliver exceptional sound and unforgettable listening experiences at home. Try it for yourself at a focal powered by name boutique. Visit focal powered by name.com for more information. So we've spoke a lot about the, I guess you describe it as the ethics for how AI interacts with humans or the copyright, those sorts of sides of it. Another part I think is quite interesting is there's this conversation around, let's say that we do start to find AI in the workforce, it does start to become a worker, quote unquote. Does it deserve to have the same kind of

workers' rights as we do? Does it deserve to have hours and lunch breaks and all this sort of stuff? Or is that where it starts to get into the ridiculous? Well, we're not. I think we have to remember that these things just computer programs. I mean, at the moment, we have businesses all over the world that run on databases that run 24-7, that run on hardware that runs 24-7. AI, in its current form, is human-like. We look at it and we see its output. And I think it's been quite a revolutionary moment for a lot of people who are not in the IT space, who are suddenly seeing all these science fiction promises of kind of robots and thinking machines, you know, suddenly they can go on to chat GPT and see something which actually looks like a human, as if they're talking to a

human. So we've kind of given the impression that this stuff has human-like intelligence. It's not sentient. There do seem to be some emergent reasoning capabilities, which I think are a little bit unexpected. But that's really, you know, that doesn't, that's not a sentient machine. It can't feel, it can't think, which is probably fortunate. But I think at this point, no. I think we should think of AI in its current form in exactly the same way we would think about a database or would think about any other computer program.

If, and then there's a very, very big if here, if we reach a point where we do, manage to create general machine intelligence, and that general machine intelligence does have human-like characteristics, or even let's not get, let's not even take it up to human level. Let's say it's, we create an AI which has the same level of, quote, intelligence and understanding as a donkey. So, don't know why I chose a donkey, but let's think about a donkey. The fact is at the moment, if you use a donkey to transport things around the place, you're not allowed to overwork it, you have to feed it and stable it properly, we give rights to things

which are in some minimal way sentient. We give rights to things which are capable of being tied. We give rights to things which can be overstretched and exploited. So I don't think anyone's ruling out the thought that maybe we will begin to create models which are more biologically inspired. I mean, I know of a lot of researchers in the field who are looking to biology in terms of how they generate the next wave of neural networks. And then we call them neural networks because we're trying to mimic the way the brain works. So if we generate things which genuinely kind of have, you know, animal-like capabilities, let alone human-type capabilities, then yes, there may come a point where we need to start granting them rights.

And that is going to cause massive uproar. I mean, just the religious element to it. I mean, the thought of suddenly saying, we've almost acted in a godlike way to generate these organisms from nothing. It's like, you know, genetically all generated neural networks. I think that we're going to have the person who does that is going to be probably the most famous and also the most pilloried person in the history of humankind. Obviously, I mean, as you say, right now and unlikely for a long time, AI can learn a

compassionate outlook and it can understand what that is, but it can never actually experience it. And I wonder if that then becomes an issue if it tries to be implemented into areas where compassion is more important. I guess I'm thinking in places where it is already being suggested in places like the legal or the health system. Is it fair to have a model or a worker that can't actually understand compassion but just knows what it is? Let me phrase the thought slightly differently. I mean, at the moment, we don't try and weed out people who lack empathy from those types of profession. You know, we don't have a sociopath test where we say, well, you know, Nigel failed the sociopath test. He doesn't have empathy to any particular degree. Therefore,

should not be allowed to interact with patients, should not be allowed to interact with clients. The fact is, we as humans are very good at faking. sincerity and faking empathy, even at those times we don't feel it. And that's, you know, we need to be able to do that, really, because in our day-to-day lives, we don't like everyone we meet. We don't like everyone we interact with. And yet, we still have to provide the same level of courtesy and professional service to them that we would if we did like them. So, that idea of fakery is a requirement for almost every job at the moment. So I don't think we could or should exclude a machine that's capable of doing it.

And frankly, is better at doing it than we are. And we're already seeing studies which suggest that when you interact with a machine rather than a human in a very conversational style, that actually people prefer the interaction with the machine. rather than the human. It's more consistent. It's generally more accurate. It's only when we get into the slightly more difficult, ethical questions that we ask of things,

that it becomes a little bit trickier. Would I ask a machine whether or not it should switch off a life support system? No, I absolutely wouldn't. But would I ask it to help diagnose my cancer? then yes, I absolutely would. So I think if we as humans are sensible about the extent to which we put these machines in front of other people, then that's a good thing. I mean, just thinking about the, we could revolutionise healthcare by effectively giving every single person on the planet access to a doctor, at least someone who could triage problems straight away.

We could do that using this. And we can do it in a way which is empathetic. So I'm very, very much in favour of pursuing those types of goals with AI. And yeah, let's allow a machine to put a human face on and give the level of service that people expect. We're currently in a massive AI boom. I mean, as this whole conversation has proven, it seems to be all around us right now. There's innovative new products launching around us. How do you see the discussion of ethics in this realm unfolding in the next couple of years?

Do you think, as you were just saying about, like, will there be all this regulation, will there be labour laws, or do we need to sort of take a step back and watch as it unfolds and really take the time to consider what we do next? Well, I think there's a few hidden sides of it that we really need to think about. quite carefully because we talked earlier about the fact that these models require a lot of training data to be effective and they do require a lot of training data but they also require human input as well and I think this is this is one of the things that's been overlooked in this that in order to have a model because a large all a large language model does it or all the kind of generative AI does

at the moment is it just predicts things based on what it's seen before so there's a and people have talked about them as being giant parrots. And to a degree, that's true. What Open AI did, and what Open AI did that was revolutionary is they generated a model which was able to act on instruction. So rather than just predict the next word or the next sentence or the next paragraph of something based on input that had been seen before, they gave it the ability to follow instructions so you could actually effectively ask questions of data and get it to do things.

Now, as part of that instruction generation, they had to have people review content because it was important to understand the type of content that was going in. They couldn't have racist material being produced. They couldn't have homophobic material, whatever it might be. And so they employed, they outsource this to a company who specialise in using workers in Africa, particularly in Kenya, to label this data. And so this data was put in front of people. And they were subjected to some of the most disgusting and vile text and imagery that is available on the internet. You know, bestiality, pedophilia, all of these things. and for something like a dollar an hour, they were told to go and label this stuff.

And that for me is one of the hidden sides of it. Because when we're looking at the ethics of using AI, am I really that worried about whether or not the internet's been scraped for this data? Not particularly. If we use Google today, when I type something into Google, the only reason I get search results is because that data has been great from the internet. So if you want your material to be found on the internet, it has to be available to Google. So on the one hand saying, yes, I want my stuff crawled for Google, but,

oh, no, it can't be used to train a large language model. It's a bit disingenuous, to be honest. But what does concern me is the way in which we are training these models. So when we're using humans to do this stuff, when we're asking them to label this data, there should be informed consent. And I fear that, you know, if you are a worker in Africa working on a piecework basis, you probably aren't in a position to say, no, actually, I don't really want to do this. So we have to look at the human side of training in this. And as things move forward, from an ethical perspective, we do need to look at the ethical side of implementation of it. we discussed right at the beginning about this idea of whether humans should be replaced by

AI or whether they should be enhanced by AI. I think that is, you know, that's an ethical question. You know, I know as a technologist that enhancing human behavior using AI is going to be of a lot more benefit to the workforce than just randomly dumping people. And so I would quite, you know, if I were looking at things we might regulate or not, maybe one of the things we should be thinking about is exactly that, that you place a duty on a company before it starts to use AI to say, what is the impact on my human workers? Should my first duty be to enhance the productivity of my existing workers using AI rather than randomly dumping them? Those are the types of questions that a, you know, a company should be asking themselves? Should I be giving my

customer's personal data to a company that I know very little about? You know, that's a legal question, but it's an ethical one as well. Should I keep my data within a European context? Should I use an SME to do my processing? Because actually, I don't particularly want to boost Google's profits. So I think there are a lot of ethical questions which are going to come out. And I don't think we've even begun to scratch the surface of what some of those look like yet. So, yes, we need to see it unfold, but I think that as each one of these things comes up, we need to jump on them quickly and start to ask these questions of ourselves. None of us really thought about the effect of the internet. None of us really thought about the effect of signing up for myriad free services.

we've already sacrificed our data and our privacy to companies like Google in return for free stuff. We've got, you know, we had Hotmail, then we had Gmail, we had Facebook, we've got Insta, we got all of these things where we've sacrificed data as payment for free services. I think this time round, maybe we should start asking ourselves those questions a little bit earlier in the process. You know, we should have more public debate about where is my data going? Who is using that data? Who is benefiting from that data? And this is why I think government does actually have a role to play. I think government could be a lot, you know, leaning into this a lot more and saying, actually, we are going to help fund more of this research. We are going to help protect this data. We're going to look at more secure ways of processing this data such that, cloud vendors don't get access to it. There are ways of doing it. There are technologies which allow

the inferencing of data. But of course, if we don't give our data to Google or to Facebook or whoever, then we're also going to have to think about whether we should be paying for these services. And that's a big one in itself. Maybe that's the biggest ethical question. Should we actually pay for these services? And the, you know, the quid, you know, the quid, you know, the quid pro quo for that, is we get privacy from it. So yeah, I think there's a lot of questions which are going to unfold over the next few years without question. Thank you for listening to this episode of Instant Genius. That was Nigel Canning's talking about the ethics of artificial intelligence. The Instant Genius podcast is brought to you by the team

behind BBC Science Focus magazine, which you can find on sale now in supermarkets and news agents, as well as on your preferred app store. Alternatively, you can come and find us online at sciencefocus.com. This podcast is sponsored by Name, Audio and Focal. The texture and emotional depth of music can be lost through digital sources or poor signal. Name Audio believes you can have digital precision with analog warmth. Alongside French acoustic specialist Focal, name creates high-end audio systems combining innovation with craftsmanship,

so you can listen to music. Just as the artist intended. Discover more at name audio.com. Relax and let Ralph's delivery handle your grocery shopping this week. We start with only the freshest items, then review your list and carefully choose each one. Then we pack it all up and deliver it in as little as 30 minutes, so you can feel confident it's what you ordered.

Fresh groceries, your way, with Ralph's delivery and pickup. And right now, you can save it. $20 on your first delivery or pickup order. Ralph's, fresh for everyone. You can't reason with the sun. Trust us. We've tried. This summer, it's time to put that angry ball of fire on mute.

Columbia's Omnishade technology is engineered to protect you from the sun's harsh rays that can burn and damage your skin. The sun is relentless, but so is our gear. Level up your summer at Columbia.com to spend more time outside and less time slathering on allolotion. You're welcome. Columbia. Engineered for whatever.