<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Eyal Shavit on Medium]]></title>
        <description><![CDATA[Stories by Eyal Shavit on Medium]]></description>
        <link>https://medium.com/@eyalshavit?source=rss-42ee41bde707------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*QTvYg_x-FFpmoQGEo54_oQ.jpeg</url>
            <title>Stories by Eyal Shavit on Medium</title>
            <link>https://medium.com/@eyalshavit?source=rss-42ee41bde707------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 06:00:33 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@eyalshavit/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[A Mongolian Tapestry: Landscapes, Animals, People and Their Stories]]></title>
            <link>https://medium.com/@eyalshavit/mongolia-eagles-hunters-nomads-a-tapestry-2cf6d54752bf?source=rss-42ee41bde707------2</link>
            <guid isPermaLink="false">https://medium.com/p/2cf6d54752bf</guid>
            <category><![CDATA[adventure]]></category>
            <category><![CDATA[photography]]></category>
            <category><![CDATA[nomad]]></category>
            <category><![CDATA[mongolia]]></category>
            <category><![CDATA[travel]]></category>
            <dc:creator><![CDATA[Eyal Shavit]]></dc:creator>
            <pubDate>Sat, 26 Oct 2024 15:42:04 GMT</pubDate>
            <atom:updated>2024-10-27T07:59:44.606Z</atom:updated>
            <content:encoded><![CDATA[<p>For centuries, eagle hunting was the exclusive domain of men, a tradition passed down from father to son. But in 2014, a 13-year-old girl named Aisholpan Nurgaiv shattered this gender barrier, becoming the first female eagle hunter in twelve generations of her family. It had all the makings of a Hollywood blockbuster — a fearless young girl, a traditional yet supportive family, and a gold medal at a prestigious eagle hunting competition. Aisholpan was not only the youngest competitor at that event but also the first woman ever to enter — and win — such a contest. Her story was my introduction to Mongolia, a land I had never really considered visiting, and for me, it turned out to defy expectations</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XBix2uZdr9vFcODN" /><figcaption>Little House on the Prairie, Ölgii Mongolia</figcaption></figure><p>One of the first things you notice in Mongolia is the sheer scale of everything. The land seems to stretch out endlessly. It boggles the mind and challenges your sense of scale. Yet, despite its size — 75 times larger than Israel — it has 200 times fewer people per square kilometer! Just imagine you were on a Tel Aviv beach, surrounded by 450 sun-seekers jostling for space. If this was Mongolia, you’d only see two people around you.</p><p>But there is something else you soon discover; a tapestry of vast landscapes, animals, and people with intimate stories that are framed and magnified by the vistas that surround them. This is a place where the iconic two-humped camels roam the desert, and play a starring role in the seasonal nomad migration — as a means of transportation, of course. It’s a place where large-horned goats lazily graze on endless grasslands. They provide nomads with milk, meat, and cashmere, not to mention “taking part” in the social game of “Kokpar,”. This is a form of horseback sport, a test of strength and agility that involves teams of horsemen battling to claim a goat carcass. In the west, short stocky wild horses walk between steep valleys, narrow trails, and towering mountains, setting a dramatic stage for the Kazakh eagle hunters.</p><h3>Eagle Hunter Country</h3><p>Khairatkkhan Sernahan’s face seems to have been tailor-made for portraits, so it was not hard to capture a dramatic photograph of him against the backdrop of the Altai mountains. Nested somewhere in the west of Mongolia near Ölgii, his home is an unassuming brick structure. But climb a nearby hill, and suddenly you feel the full impact of the scenery; undulating terrain stretching into the distance, dotted rugged hills intersected by winding rivers, dramatic skies filled with swirling clouds that are painted in hues of gold and blue. The scene exudes a sense of solitude and grandeur.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BVI6RhaYeDeClTyMRj2eUQ.jpeg" /><figcaption>Khairatkkhan Sernahan, Eagle Hunter, Ölgii</figcaption></figure><p>In Kazakh eagle hunting, the eagle takes center stage with its impressive wingspan — often exceeding 2 meters — keen eyesight, and powerful talons. It is not clear how eagle hunting started, or frankly why, but it is an old tradition. Archaeological evidence suggests it was practiced by ancient nomadic tribes in Central Asia as early as the 3rd or 4th century B.C. It isn’t just a skill either; it’s a way of life woven into the identity of the Kazakh people. Passed down through generations, it represents more than the hunt itself. For many, it is a symbol of the deep connection between humans and nature, a partnership where respect for the eagle’s independence is as important as its ability to hunt.</p><p>Khairatkkhan was training a recently captured female Eagle. Every morning, they would set out for a ride, the eagle perched firmly on his arm. It gets heavy after a while; so to support the bird’s substantial weight — up to seven kilograms — he used a low-tech solution: a leaning Stick (<em>Baldakh</em>) wedged between arm and leg. It supports the hunter’s arm, making it easier to carry the eagle for extended periods on horseback. Hunters have other low-tech solutions like the hood made of leather that covers the Eagles head during training and hunting. The hood keeps the eagle calm and prevents overstimulation. When it’s time to hunt, the hunter gently removes the hood, allowing the eagle’s sharp vision to lock onto potential prey.</p><p>Hunters exclusively use female eagles, prized for their larger size and fiercer temperament. While a single eagle suffices for nabbing hares or foxes, it takes a pair to bring down the ultimate prize — a Mongolian wolf. This formidable catch, though highly sought after, poses significant dangers to the birds.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Oxyd-RZoFcPhjD-8" /></figure><p>Eagles are captured from their nests at about 3–4 months old when they’re old enough to survive but still young enough to be trained. This partnership they are thrown into usually lasts for about seven years, after which the eagles are released back into the wild, often during a ceremony that honors the bird’s service.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MbbfRPCoIimOnWub49e4wg.jpeg" /></figure><p>Hunt days start at dawn, braving long hours in the rugged mountains. The hunters are wrapped in a deel. This is a garment that resembles a long, robe-like coat but it’s more than just a practical garment — it’s a symbol of Mongolian history and resilience. Worn for centuries, it has evolved with the seasons and social status, and its colors carry deep: blue, for example, represents the sky, eternity, and loyalty, while red signifies fire, strength, bravery, and love. With all of that, it’s still not a great look, but you’re in the mountains with not many people around…and it’s warm, really warm, like wearing a down blanket, and nothing beats that on a long cold day on horseback.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*m1yHXVwieofAUG_Z" /></figure><p>The hunt itself is spectacular. Eagle hunters usually hunt in pairs. One climbs a peak to hold and release the eagle, while the other helps flush out prey from hiding spots in the terrain underneath. Once the eagle sees the prey, she is released and drops quickly for the kill using her talons to grab and crush it. Hunters have to rush to the eagle to coax it away from the prey with rabbit meat they carry in their side pouch. It is, after all, a wild bird of prey.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/869/0*LqfslUFTan1eX16X" /></figure><p>There is a conservation story that is told about the practice of capturing and releasing eagles from the wild. It is an example of a balance between conservation and human tradition. Young female Eagles that would have a hard time finding food in the wild, are fed during their years of service, grow strong, and are then released back to breed naturally in the wild. This is good for managing the eagle population. It may seem more brittle from the Eagle’s perspective. Tied with a plain rope and wearing a head cover leather cap most of the day these creatures are also wild animals held in captivity. It takes a while to get used to this contradiction and it certainly helps to place it in the context of 4000 years of tradition passed through the generations; father to son and now also father to daughter.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1tyJVGzEP_z2sdgeflvdnw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nlTZQVLinXoW2B2y" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7o-tkLcsWJKWmXa2S-hEUg.jpeg" /></figure><p>One early morning, I saw Khairatkkhan taking a horse ride through the pastureland surrounding his home, his eagle perched on his arm. He was dressed in full hunting regalia, ready for battle, which I found an intriguing choice for such a short excursion. Curious, I asked him why all the fuss. His response, more of a shrug really, was that this is a tradition that has persisted for centuries — a vital process through which eagles and their hunters acclimate to one another. This he said, was as much for him as for the eagle, it was part of their journey together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3jmL0xqsEG7w3rnR" /></figure><h3>Nomad Country</h3><p>To the south, in Khovd, near the Mongolian White River, Sükhbat and Enkhtuya are the Aav (Father) and Eej (Mother) of a Nomad camp group that live and work together. This is a group of families, not necessarily related, that live in close proximity and form a small community. They work, socialize, and migrate together. Nomads migrate to lower elevations and protected valleys in the winter and to higher elevations and wide-open pastures in the summer. This migration is not just a practical move — it’s the heartbeat of nomadic life. Every shift in location is part of a cycle that ensures the survival of both the people and their animals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*pjyWx05SHZTaIi53" /></figure><p>At first, Sükhbat and Enkhtuya were reserved, this was one of the first times this group hosted outsiders. But soon enough banter took over, quick-witted and entertaining. Sükhbat had not heard of Israel but had heard of the Dead Sea and its powerful minerals. He wanted to know how much it would cost to send Enkhtuya to the Dead Sea to keep her young. She was quick to point out in front of everyone that it may not be in his best interest, because, well if the minerals did in fact work, she is not certain she would come back. Yes, he nodded, with a slight smile, maybe it is better if I go to check it out myself first.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/578/0*zMlhmWiItrfb55Yq" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/736/0*F9_8zpo3dekEpTIz" /></figure><p>They reminded me of a big Italian family straight out of Ettore Scola’s movie “La Famiglia”, with cousins you feel too close to and relatives that you’re not quite sure who they are or how they are related to you. It’s a scene brimming with life and laughter as they manage their expansive herds of goats, horses, and camels.</p><p>Goats in this region are often Cashmere goats, but to me, these did not look like goats that could make a fashionable sweater. They seemed strong, hardened, and well adapted to the elements with thick rough coats and strong pointy shaped horns.</p><p>The day for the goats begins with herders preparing for milking. They are gathered and tied in a “zipper formation” to make the process efficient. The herders use the word “Kholboo,” meaning ‘connection’ or ‘line,’ to organize the goats. Alternate goats face in opposite directions and thus are bound together in a manageable milking line. Most goats seem familiar with the routine and maybe even welcome it; it gets cold in the Mongolian plains and Kholboo provides warmth. But there is always one or two that manage to escape, adding a touch of humor to the routine, as goats are chased around by herders and brought back into the line. After milking, , sometimes wet goat dung is applied to the teats to prevent baby goats from going around nursing all day. This helps not only to manage the milk supply for the group but also makes sure not to exhaust mama goats.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Xehqgc_tia9_xX5u" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/696/0*VPEanST3NRYorrA1" /></figure><p>Goats have a lot of free time. They spend the day grazing in different pastures so adding a touch of paint to their horns is a low-tech way to easily identify the goats that belong to your herd as they roam about.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*xZ-MiQBlV4rVk1M-" /></figure><p>Horses here are legends in their own right, carrying warriors like Genghis Khan across the once vast Mongolian empire. They are tough and wild animals that are unlike the domesticated horses we know — they don&#39;t live in a stable and remain indifferent when you try to ride them. They are more like smaller stockier horses with an attitude. They get released in the winter to fend for themselves and are recaptured in the summer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1022/0*yl2kXmGPA7hOivHk" /></figure><p>They have a large head, thick neck, short legs, and a dun-colored coat with a dark mane that stands erect. They know how to find food, digging through the snow with their hooves to reach the grass underneath (a behavior known as “Tovokh”). They also seem to move when they want to, and not necessarily when you tell them to. In the mountains with the Eagle hunters, I found myself more than once on a narrow path with sheer cliffs and daunting drops, trying to convince the horse that it was time to catch up to the hunters who were quickly moving through the terrain. One of those times, apparently sensing my discomfort, Khairatkkhan looked back and raised his hand wiggling 4 fingers at me, as if to say, trust it, it has 4 legs, it won&#39;t fall, you will be just fine. I don’t remember it being that comforting…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*NjLgTasxp0-6Yqr-" /></figure><p>While the Eagle hunters we met seemed serious and solitary, Nomads, we soon realized, were highly social. They devoted as much time to play as they did to work. They pranked each other constantly, playing games and competing with one another. Two of the young horse herders were very eager to show off their riding skills, picking up objects mid-gallop, showcasing their agility and precision at controlling their horse. They were at it for a while, us admiring their skills, the other nomads making fun of them and their youthful bragging….. until at one point Narangerel (a woman’s name that translates to sunlight) had had enough: she got on a horse, did a perfect head-spinning run, and calmly walked away smiling as if “dropping the mic”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ioAq2_fzIfkpfjXm6jJQ1g.jpeg" /></figure><p>Camels are the stars of the “migration party”, where in the winter Nomads move to to lower elevations and protected valleys, mainly in a “transportation” role. Migration means packing, and it’s an all-hands-on-deck type situation, with Sükhbat and Enkhtuya firmly running the show.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Y804L8Z1waAhePYd" /></figure><p>Within a couple of hours, the Gers — traditional, portable, round tents used by nomads — are dismantled piece by piece. Equipment is carefully bundled, and camels are guided to lie down so the loads can be mounted and secured for the long walk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FFsASpdPw2ODaJhH5IQSqA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Dia2OVIZEMoTdvXI" /><figcaption>left: Ger the traditional nomad dwelling. Right: Camels packed during the winter migration</figcaption></figure><p>What follows is a flurry of activity: straps and ropes are pulled and tightened with precision. The process takes longer than I had anticipated, filled with the sounds of people shouting instructions and camels grunting in protest and making a general fuss. Observing my bewilderment, Sükhbat explains that this seemingly straightforward task is crucial. Properly securing the equipment can mean the difference between a smooth migration and a challenging journey</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ucyPR3Gbdt59CFE7DB3iDQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0HKJsTYxjBBAVqij" /><figcaption><em>The migration — majestic, walking through ravines and mountains</em></figcaption></figure><p>The Story of Aisholpan Nurgaiv and her family, first discovered by the National Geographic Photographer <a href="https://www.svidensky.com/project/the-eagle-huntress"><strong><em>Asher Svidensky</em></strong></a>, was made into a remarkable <a href="https://www.imdb.com/title/tt3882074/"><strong><em>movie</em></strong></a>. Its dramatic plot captured my imagination before I ever set foot in Mongolia. Being there, the reality of experiencing its people, their land, and their stories turned out to be a much more visceral and authentic experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/378/0*EUGpjRx3iyJ1gpMK" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oCYlTw4FOR5nTeEI0y2BIA.jpeg" /><figcaption>“I am Mother” , Ulma, Aisholpan mother</figcaption></figure><p>As I journeyed to Mongolia, a passenger in an adjacent seat asked me “Why are you going to Mongolia? I mean there’s <strong><em>nothing</em></strong> there”, and at the time, I didn’t have a compelling answer. Mongolia wasn’t a place that I had planned on visiting. It was a destination that spoke to me unexpectedly through a story about eagle hunters. What I discovered though was far from “<strong><em>nothing</em></strong>”. I found a rich tapestry of landscapes, animals, and people. I found not emptiness but vibrant human stories full of humor and wisdom that are inextricably linked with the natural world. I found an experience that celebrates the profound beauty of “<strong><em>everything”</em></strong>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2cf6d54752bf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Rise of AI-Products]]></title>
            <link>https://medium.com/@eyalshavit/design-re-thinking-the-rise-of-the-ai-products-9644e7f862b7?source=rss-42ee41bde707------2</link>
            <guid isPermaLink="false">https://medium.com/p/9644e7f862b7</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-products]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Eyal Shavit]]></dc:creator>
            <pubDate>Tue, 30 May 2023 11:20:29 GMT</pubDate>
            <atom:updated>2023-06-01T10:15:33.239Z</atom:updated>
            <content:encoded><![CDATA[<p>Tomorrow’s products look nothing like yesterday’s. AI is everywhere and it is giving rise to a new category of “AI-Products”. AI is more than a feature in these products — it drives their core. This makes possible completely novel experiences for users; AI enables Netflix recommendations today, sure, but it also makes possible prosthetics that will adapt to your gate, cars that will detect and respond to how (they think) you <a href="http://venturebeat.com/ai/affectiva-launches-emotion-tracking-ai-for-drivers-in-autonomous-vehicles/">feel</a>, or personalized food that is 3D printed for you on-demand.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3VjOlLSCTcjQtAXEahWlbQ.png" /><figcaption>Created using Midjourney</figcaption></figure><p>Today we think about AI as working behind the scene, “good AI is transparent AI”, but as we drive it deeper into our products, this will no longer works. We believe that what makes AI-Products into a category of their own is…well…the AI; you depend on it to deliver value, it is constantly hungry for data, and the way you interact with it is sometimes not so obvious. For example, to best interact with ChatGPT you need ‘Prompt Engineering’, there’s a TikTok channel about ‘Alexa Gone Wrong’ because of bad data, and if you take the AI out of an autonomous car, you are not getting a not-so-great autonomous car…you are getting just a car. <strong><em>So as AI continues to advance and even accelerate, how do we properly design these products?</em></strong></p><blockquote>We argue that AI-Products need an “AI-Native” design framework as today’s frameworks were born pre-AI</blockquote><p>When compared to traditional products, AI-Products can engage users in entirely different ways (<a href="http://www.youtube.com/watch?v=YvT_gqs5ETk">SNL’s Echo for Elders</a> is a funny take), but at the same time, they are also more complex and introduce a new set of challenges for designers. Today’s design frameworks, like Design Thinking or User-Centered Design, were not designed with AI in mind and therefore lack some of what is needed to handle this complexity. For example</p><ul><li>Traditional frameworks drive to define the experience and all interactions up front, whereas the behavior of AI-Products is dynamic and adaptive and hard to completely define upfront.</li><li>On the data side, product behavior today is designed from small datasets that are gathered at the beginning of the the design process (e.g. interviews, surveys). AI-Products’ behavior relies on large amounts of data that are collected in real time and used to drive learning and adaptation.</li><li>The AI in AI-Products can result in non-obvious behaviors that are tricky to specify in advance as they are situational; they can arise quite clearly under one set of circumstances yet not at all in (what we think of as) very similar circumstances. This is a challenge for traditional design frameworks.</li></ul><p>We motivate and describe an <strong>AI-Product Design Framework</strong> that we developed and have been using for a few years now. We will start by defining AI Products and exploring their characteristics. We’ll then talk about the new challenges they pose and what it means to design them. This is all part of a much larger conversation about AI and Design; Generative AI is already transforming designer tools and workflows, but it is also beginning to change the very role of a designer. <strong><em>Our longer-term goal is to bring key issues that are at the intersection of AI and Design and develop a practical framework to design AI-Products.</em></strong></p><h3>Understanding AI Products</h3><p>AI-Products have artificial intelligence at their core, driving the primary functionality and value proposition for users. Perhaps the simplest example is the new generation of thermostats; a traditional thermostat maintains the temperature that is set by a user through some form of physical control (a button, a dial). The new generation of thermostats, like the <a href="https://store.google.com/us/product/nest_learning_thermostat_3rd_gen?hl=en-US">Nest Learning Thermostat</a>, adapts using AI, automatically adjusting temperatures based on learned schedules, and even weather conditions. One provides a static experience that is fully controlled while the other is a dynamic one that can deeply integrate with a user’s habits, routines, and lifestyle.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*CABaB_FMCwjOrBuH-7eD4w.jpeg" /><figcaption>Rendition of a Traditional thermostat (left) vs the new generation of Nest Learning Thermostat (right)</figcaption></figure><p>One of the things that characterize AI products is that they constantly operate to understand Context. In AI, context refers to the understanding and use of situational information and environmental cues to help the AI make sense of its interactions. This is crucial because it allows the product to deliver more relevant, highly personalized, and intuitive experiences to users. How to design these experiences however is still very much an open question; they move us from “Click”, “Press” or “Select” type interfaces to a natural user interface (NUI) that mimics the way that humans interact with the world around them (e.g. through gestures, voice, conversation, or vision). <em>AI products depend on the technical quality of their AI, but equally, if not more so, they depend on the caliber of their design that defines the what, where and when of applying the AI.</em></p><h3>So What is an AI Product?</h3><p>AI-Products have several distinct characteristics, which make them unique</p><ul><li><strong>AI is at their core</strong>. The product must fundamentally rely on AI for its functionality and its value propositions. If you take the AI out of an autonomous car, you are not getting a not-so-good autonomous car…you are getting just a car.</li><li><strong>They operate to understand Context</strong>. They go beyond basic reactions to changes and try to understand the circumstances surrounding those changes. Smart Home Security Systems use AI to differentiate between usual and unusual activity in your home. If the system is familiar with your daily routine, it may recognize a movement in your house at 4 AM to be unusual and alert you or take appropriate action based on its understanding of the context.</li><li><strong>They Adapt.</strong> The product must be able to learn and adapt to its environment and improve its performance over time. This is based on data that is continuously collected about users and their interactions. <a href="https://www.waze.com/apps/">Wayz</a> for example will adjust route recommendations based not only on where you want to drive to, but also on real-time traffic conditions, social traffic reports, the weather, and past your drives.</li><li><strong>They make autonomous decisions</strong>. AI-Products can function in their intended environment, making real-time decisions, without human intervention. A smart home lighting system can learn your daily routine and autonomously adjust the lighting depending on if you are cooking, watching TV, or reading a book, even before you realize you need it.</li><li><strong>They are highly personalized</strong>. Devices like Amazon’s Echo or Google Home’s voice command recognize individual voices from different members of a household. They learn a user’s preferences over time, including their music preferences, favored news sources, preferred shopping items, etc. Smart appliances, like Samsung’s smart refrigerator, make eating easy for you — it gives you recipe ideas based on your food likes and dislikes, helps you make a meal plan for the week based on what’s already in your fridge, and even creates a shopping list with all the ingredients you’ll need.</li><li><strong>They have a Natural User interface</strong>. Mimicking human interaction, through conversation or gesture, makes an interface more accessible, easier to learn, more personal, and more engaging. <a href="https://www.livongo.com/">Livongo</a> has products that help patients manage their chronic conditions. Their conversational interface is personalized to individual users, providing “Health Nudges” just at the right time. This not only improves long-term engagement but also increases a patient’s trust in the product.</li></ul><p>Let’s look at a hypothetical AI-powered robot caregiver as an example of an AI-Product. The core functionality of this robot caregiver lies in its AI. It enables the robot to understand, interpret, and respond to the physical and emotional needs of its user. It does so by collecting and interpreting data from its sensors, making decisions based on this data and what it learns from interacting with the user over time. Its interface is natural and human-like. It uses speech recognition and synthesis for voice-based communication, computer vision to recognize faces and gestures and touch sensors to respond to physical contact. A user might verbally ask the robot to fetch a glass of water, and the robot would understand the request, find the kitchen, locate a glass, and fill it with water. The robot caregiver adapts over time as it learns and adapts to the user’s routines, preferences, and needs. It might learn when the user typically wakes up and goes to bed, their preferred indoor temperature, and the kind of TV shows they like to watch. It uses this information to provide better, more personalized care. The robot caregiver has a degree of autonomy and can function without human intervention. It can navigate around the house, remind a user to administer medication, detect the room’s temperature, and adjust it if it is too low. It can also adapt to a user’s conversation style or suggest meals to personalize the user’s experience. The robot caregiver will understand context, responding to a midnight request for a glass of water, by moving quietly to avoid waking others in the house. In this example, the robot caregiver, an AI Product, integrates AI technologies and provides a service that is personalized and responsive to the user’s needs and context, demonstrating the unique potential of these products.</p><h3>What are the challenges in designing AI Products?</h3><p>A fundamental paradigm shift for designers is that <em>AI-Product’s behavior depends on continuous real-time data for normal operation</em>. The AI drives the technical need for data, but from a design perspective, the entire behavior depends on it. An AI-driven cooking app, for example, provides real-time feedback and adjustments to the user while they cook, based on inputs they collect from smart kitchen appliances. A Chatbot depends on real-time responses when you have a conversation with it. This requires a change in how we handle real-time data in our current design processes.</p><p>Interpreting the results of the AI is another challenge. Think of it as an <em>“error-handling” issue</em>; errors in AI Products are not just from software bugs but can emerge in complex ways. How does a product handle the AI not working at all? what is the right action when the AI partially fails and gives results that are strange but not obviously wrong? How do we handle situations when users do not interact with the AI in the ways designers anticipate, provide unexpected input, or misunderstand AI’s output? Understanding how the product should behave in these cases can be tricky, and designing for them requires careful consideration and clear communication to reduce user frustration.</p><p>Related is the topic of <em>Transparency and Explainability</em> which we think will increasingly draw our attention. It’s important for users to understand how and why the AI is making certain decisions or recommendations. A doctor needs to understand how an AI-driven diagnostic device reached its conclusions. A patient needs to understand why were they diagnosed, with what exactly, and how worried they should be. Transparency and explainability do all kinds of good things; decrease user frustration, increase trust, and make sure products are ethical and don’t propagate bias.</p><h3>Designing AI Products</h3><p>While we are still very early in the development of methodologies to build AI products, it is already clear that current design frameworks are not sufficient and need to evolve. This is not surprising as traditional design frameworks, like <a href="https://en.wikipedia.org/wiki/Design_thinking">Design Thinking </a>and <a href="https://en.wikipedia.org/wiki/User-centered_design">User-Centered Design</a>, were not constructed with AI in mind. To us, a design framework that is “AI-Native” requires the following ingredients</p><ul><li><strong>Increases a designer’s understanding of AI</strong>. Designers need to have a clear understanding of AI’s capabilities, limitations, and implications. They don’t need to know how to code their own AI systems but rather understand the basics of AI and the data that is gathered and used. This is an important part of making informed design decisions.</li><li><strong>Focuses on an “AI Native” action model</strong>: Traditional design frameworks focus on a User-Action model where we define users, needs, and possible interactions upfront. In an AI-native model, this is extended to a User-Data-Action paradigm. Users are dynamic, actions are dynamic and data is recorded and used in real-time to connect them throughout the product lifecycle.</li><li><strong>Design for Adaptability</strong>: A distinctive aspect of AI products is their ability to learn, evolve, and adapt to changes over time. These adaptations are shaped by the interactions and responses of users, creating a dynamic feedback loop that continually refines the product. Designers need to adopt a more fluid approach to design. They need a framework that allows them to anticipate and design for a range of possible future states, rather than a fixed set of predefined interactions, to plan for user experiences that may vary greatly from one user to another, or even for the same user at different times.</li><li><strong>Design for a Native User Interface</strong> (NUI): Interacting with a product through voice or gesture is a departure from today’s popular interaction model that often revolves around a combination of graphical user interfaces (GUIs). The goal of AI Products is to make the interaction as natural and human-like as possible, almost like a conversation you would have with another person.</li><li><strong>Drives for Simulations not just prototypes</strong>: AI products have dynamic users and dynamic behaviors that interact through a NUI. While prototyping is already a part of the design process, simulation is not, and it is needed to account for the dynamic nature of these products.</li><li><strong>Promote and Enable Collaboration: </strong>traditional frameworks have prespecified points in the process to gather stakeholder feedback. Designing AI products requires close and continuous collaboration between designers, AI experts, data scientists, and other stakeholders. This type of collaboration is critical as it fosters knowledge exchange, helps understand AI capabilities and limitations, and can address ethical concerns in the right forum.</li><li><strong>Promote Explainability</strong>: AI Products need to clearly articulate their actions or decisions. In non-AI products, explainability often comes down to usability design, a menu of similar actions, a help tooltip, instructions, or manuals. In AI products, however, explainability becomes much more challenging as AI decisions are not hard-coded and may even appear to be counterintuitive or incorrect to a human observer. Designing with explainability in mind is directly correlated with the perceived quality of the experience for the user and the level of trust they have in the product.</li></ul><h3>An AI-Centered Design Framework</h3><p>It comes as no surprise that, as progress is typically an evolution and not a revolution, we build on proven elements of pre-AI design frameworks. We adapt and extend some, and add new ones that address specific needs of AI Products. At a high level, the AI-Centered process can be articulated as a set of phases</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jcoJ2WCwAsZCtx2oftXWVw.png" /></figure><ol><li><strong>Contextualize: </strong>This is a combination of the User Research and Empathize phases of UCD and “Design Thinking” respectively. We conduct quantitative user research, surveys, interviews, and user observations to understand the user and their daily needs. We also drive to understand AI-specific context, for example, understand the user’s experiences with AI; their level of trust, privacy concerns, and comfort level. We typically distill our insights and finding into a <a href="https://pdmethods.com/pain-gain-map/">pain-gain map</a>.</li><li><strong>Evaluate: </strong>where we distill many of the observations to clearly define the value(s) for each (type of) user and specifically how AI can deliver this value. It’s critical that the value is not simply to “use AI” but rather to address a real user need that AI can satisfy. This should directly address the pains and the gains that were identified in the Contextualize stage. Also in this phase, we collaborate with AI experts and stakeholders to generate a range of innovative AI solutions that address the user’s needs and challenges.</li><li><strong>Design: </strong>Here, we design the way in which the AI delivers the value to each user. This involves designing the user interfaces, the AI behavior, and the user-AI interactions. We consider the many possible states of the dynamic interaction of AI products and how data is collected and used to adapt the experience. We define how the AI will communicate its capabilities, limitations, and decisions to the user.</li><li><strong>Validate</strong>: Designers work with AI experts to validate that the desired AI solution can in fact deliver the value to users in real-world situations. We understand the AI’s needs in terms of data, computational resources, and development time. We identify what we will need to properly prototype and simulate the product’s behavior and articulate AI’s limitations such as accuracy, robustness, and explainability.</li><li><strong>Prototype &amp; Simulate:</strong> Here we will build a lightweight realization of the design and a simulation environment. We aim to answer the “can” and “how” questions. Can we realize the idea and value to users given our design, resources, expertise, and experience? And if we think the answer is, “Yes, we can”, then we need to answer the follow-up how questions about what do we need to implement, what type of AI will work, and how users will interact with it to get the value. Given the adaptive nature of AI, simulations are critical and need to consider different user contexts and behaviors over time. For instance, a prototype of an AI chatbot might include simulated conversations that show how the chatbot learns and adapts its responses based on user feedback.</li><li><strong>Iterate &amp; Refine:</strong> We use this phase to test the AI product with real users, collect feedback, and iteratively refine the product. This is similar to the test phase in traditional design thinking and UCD but also involves testing the AI behavior and how the behavior evolves over time. Once the prototype is in the hands of users, it is easier to gain insights. For example, it allows us to understand what works and what does not work in the first-time user experience. Instructions that seem obvious at design time may baffle users and require an iteration. Similarly, we may find that having no data about the user degrades the experience and we would be better off to actively query the user for information in this stage. This phase includes collecting both quantitative data (e.g., interaction metrics, model performance) and qualitative feedback (e.g., user interviews, observations).</li><li><strong>Align:</strong> this is an important phase to call out. We explore how the designed product aligns with a range of human-centric considerations, including ethical implications (e.g. fairness, discrimination), privacy concerns (e.g. user consent), or bias propagation (e.g. gender).</li></ol><p>We’ve also developed best practices and tools as part of the AI-Centered Design framework. For example, we capture the different steps of this process in a design canvas, inspired by the <a href="https://www.strategyzer.com/canvas/business-model-canvas">business model canvas</a>, which serves as a tool for designers working to design AI-Products.</p><h3>Discussion</h3><p>In this amazing era of AI, the landscape of product design is shifting dramatically. Tomorrow’s products are radically different from those of yesterday, and form a category of their own. This is primarily due to the transformational power of AI. But with this power come a set of considerations that are new and that need to be included in how we design AI Products. AI is already transforming designer tools and workflows, making it simpler to ideate and removing many of the time-consuming obstacles that designers deal with daily.</p><p>One of the less discussed impacts is a change in the role of the designer. Tomorrow’s designer needs to be both a creator and a curator, “design-literate” but also “AI-literate” and “data-literate”. They will need to develop ways to design products that evolve in unpredictable ways once they are in the hands of the user, and at the same time, they will need to plan around the tricky issues that can lead to unethical, unfair, or biased products.</p><p>In our follow-up post, we will elaborate and expand on the AI-Product framework, showing how to use it in detail and pointing to insights and practical recommendations for designers.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9644e7f862b7" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>