<?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 Dylan Hong on Medium]]></title>
        <description><![CDATA[Stories by Dylan Hong on Medium]]></description>
        <link>https://medium.com/@dylanhongtech?source=rss-457f74ce468d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*Hued47bNYCvCubRdvh2-fg.jpeg</url>
            <title>Stories by Dylan Hong on Medium</title>
            <link>https://medium.com/@dylanhongtech?source=rss-457f74ce468d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 06:21:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dylanhongtech/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[Why The Humanoid Tesla Bot Does Make Sense]]></title>
            <link>https://medium.com/@dylanhongtech/why-the-humanoid-tesla-bot-does-make-sense-ece69678dab0?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ece69678dab0</guid>
            <category><![CDATA[tesla]]></category>
            <category><![CDATA[robotics]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Mon, 30 Aug 2021 22:56:08 GMT</pubDate>
            <atom:updated>2021-09-01T14:02:38.989Z</atom:updated>
            <content:encoded><![CDATA[<p>I was recently listening to the WVFRM podcast by Marques Brownlee and Andrew Manganelli. In it they ask the question, is a humanoid robot a good idea? Essentially, the argument is that it is better to have a bunch of specialized robots rather than build a humanoid one. And is there any actual benefit to a humanoid shape compared to any other physical design?</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FZBzZ8OoPmrA%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DZBzZ8OoPmrA&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FZBzZ8OoPmrA%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/87fc2a4f554f6d1f1646e25aeb219831/href">https://medium.com/media/87fc2a4f554f6d1f1646e25aeb219831/href</a></iframe><p>I’m an HCI engineer (human computer interaction) — you could think of that as UI/UX, I’ve done work on Augmented reality experiences, and even some hand tracking and gestural controls, and I also really like robotics and autonomy. I wanted to share some thoughts on why I think the humanoid robot is actually a good idea.</p><h4><em>If you’d prefer hearing me read this article, I created </em><a href="https://www.youtube.com/watch?v=rcAG5G2Cuys&amp;t=73s"><em>this video</em></a></h4><p>In the product design space we often have to view these sort of projects with a desirability lens and then contrast that with a feasibility lens. I think in regards to Tesla’s ability to actually produce a bi-pedal robot that can do all of the things they mentioned, I’m not as optimistic as their timelines in the short term. But to explore why humanoid makes sense, I think we’ll just have to assume that they can pull it off (a humanoid robot that can see, navigate, and interact with the world in ways very similar to a human using AI). And then that lets us focus just on the humanoid form itself. I know it’s a lot to assume, but let’s roll with it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bgj5cN-Zz18VTNeKbRpqnA.png" /><figcaption>Tesla Bot Announcement Slide</figcaption></figure><p>There are three main reasons I think it makes sense to explore a humanoid form.</p><ol><li><em>Generalized solution</em></li><li><em>Increasing Theoretical Maximums/Designed for Edge Cases</em></li><li><em>Consistent UX/interaction design</em></li></ol><h3><strong>Generalized Solution</strong></h3><p>Funnily enough, Elon actually noted that a specialized solution will always accomplish a task better than a generalized solution when he was talking about Dojo. This is true, but to the human experience, many tasks that we do have diminishing returns when it comes to precision. In a factory or neural network training, having a specialized solution makes sense because you assume that we’re not at the point of diminishing returns when it comes to building a car that’s built within microns to spec — we need to keep pushing toward this. But when it comes to something like folding your laundry or opening your door and brining a package in your house, the generalized solution will probably be fine.</p><p>If you take the view of, having a hundred specialized robots or tools is better than investing energy into building the Tesla Bot, I think a good product to see how a similar debate played out in the past was with smartphones. Before smartphones we had specialized tools (or robots as you could call them) to listen to music, to give us directions, to take pictures, to send messages, and the list goes on and on. Sure, a specialized music playback device could give us a better music experience, but clearly people cared more about having their music with them at all times so we lived with the tradeoff. A traditional point and shoot camera took better photos and had zoom compared to the smartphones that replaced them, but the best camera is the one you have on you. Generalized or all-in-one solutions require so much less thought and are so much more convenient for the end user.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/0*jcnH5kxN8Efifkwr.jpg" /><figcaption>Steve Jobs Unveils iPhone</figcaption></figure><p>Additionally, in terms of capital efficiency, iPhones were very expensive when they first launched. They were more expensive than the devices they replaced and did their tasks less well. However, because Apple only had one form factor they had to worry about, they were able to design a process which allowed them to take advantage of economics of scale. They sold so many iPhones that now you can buy an iPhone for much less than what it would have cost to buy a GPS, MP3 player, Camera, web browser. With the Tesla bot, you could theoretically achieve this scale benefit much quicker than if you were to build hundreds of specialized robots. And as Apple did it, they took the cost savings and were able to reinvest into the product to make it better until there was essentially no gap between the quality of the specialized solution compared to the generalized solution.</p><h3><em>Increasing Theoretical Maximums/Designed for Edge Cases</em></h3><p>When you design a product, you come up with vision content around it. The goal here is to think about what could thing be 1 year, 3 years, 10 years down the line. If you do this experiment with a robot vacuum, you might be able to come up with some novel ideas about how the vacuum will have higher suction power, or eventually be able to lift stuff up and get underneath things, or even climb up and clean the walls. But at the end of the 10 years, the vacuum with still be just that. So when you are thinking about building this vacuum robot product, your theoretical maximum is likely focused on how clean the house can be. That is the goal you are solving toward.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R43ipj0VtISgszxzuZD9jw.png" /><figcaption>My Robot Vacuum</figcaption></figure><p>When doing the vision exercise with a humanoid robot, we can align the theoretical maximums of the Tesla bot with those of ourselves. Maybe now the bot can get groceries and open doors for us, but in 3 years maybe it will be unpack those groceries and use them to cook meals for us. Maybe in 10 years it will be able to maintain a home garden and come up with new recipes based on food science, cook and serve, then clean all the pots and pans when I’m done eating. Here, this was only one subset of Tesla bot’s potential uses. The theoretical maximum in this field extends to agriculture, nutrition, and cleaning. This is just for one use case. We don’t really know the theoretical maximum of humans because when we face physical maximums like running speed or distance, we build cars or airplanes. To align the Tesla bot to the general abilities of a human, should help expand the theoretical maximum of the product, increasing its total addressable market when then feeds into the economies of scale again.</p><p>Related to maximums are edge cases. If you design a robot to have one task like a robot vacuum, how many edge cases will it be able to solve? We know they can detect stairs and even different types of floors, and yes you can program them to not go in certain parts of the house, but what happens when there are foreign objects in its way? Or what happens where there is something delicate left on the floor by a guest that happened to drop it during the scheduled cleaning sessions? Will the robot, that is designed with the goal of cleaning, be able to understand that this object should be moved to a different location before continuing with the room? If it does know this, how does it move the delicate object safely? It was only designed to be a vacuum so it likely doesn’t have appendages that would make manipulating small objects possible. If you wanted to design a robot that could handle all of the edge cases in a human world, it needs to be able to visualize and manipulate the world like a human. This means, all of the task-purposed robots would need to have hands and legs anyway which begins to defeat the purpose of having specialized robots. While, yes, there will be edge cases that the Tesla bot is not up to the challenge for, but because it is built as a general purpose machine, it could be iterated upon though hardware or software in a way that better handles the edge cases like a human could without taking up additional space or adding further complexity to the spaces we occupy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SYlevFO36uwkFix1i7pB4Q.png" /><figcaption>Tesla Bot’s Technical Details</figcaption></figure><p>If Tesla is able to solve real world AI, and I know this is a big “if” but for the sake of the humanoid angle we have to run with it, building hardware for it that is not constrained by preset maximums will be extremely important in the goal of scaling its adoption. If you are focused on solving all of the small problems (cleaning, transport, cooking, packing), you will always be constrained by the scope of the project. We already have a design with an unknown theoretical maximum, that is humanoid.</p><h3>Consistent UX/Interaction Design</h3><p>Think of all of the the smart things you have in your home, more likely than not, they all require you to learn a new interface. They are often built by different companies, they might work with Amazon Echo but not Apple’s HomeKit. They might have a hardware remote or be controlled by an app or just voice. They might connect to the internet and stream data or they might just exist and do their job. Even something that should be simple like a tv remote, has way too many inputs when really the user only has one main goal “watch the thing I want to watch.”</p><p>Now, let’s expand this to a future where we have robotic instances of all of the appliances we have and for all of the chores we want to do. Will people have to teach their grandmothers that they can speak to the washing machine but they have to use their phone to vacuum the floor? Will the robots get in each others way if they are trying to complete tasks in the same part of the house? How many people use an iPhone every day, but get tripped up when given a task on an android phone? Having a consistent interaction paradigm for our autonomous assistants will make the technology accessible to many more people. Additionally, we already understand the human interaction paradigm. We talk or use sign language and we point or gesture. Humanoid robots would enable a human machine interface similar to that of just talking to a human. Asking your Tesla bot to clean the house twice a week while you’re at work is a lot easier than scheduling your dusting robot, vacuuming robot, window wiping robot, toilet cleaning robot, and others using a combination of phone apps, web pages, and voice commands.</p><p>We already see that Open AI is able to accomplish impressive natural language processing (GPT3), with a humanoid body and vision/audio sensors, we could also see something like natural gestural processing in the future as well.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qKLxgqzzHZDy9YYJ2SHDog.png" /></figure><h3>Conclusion</h3><p>These were the three main reasons that I’ve been thinking about when it comes to why, if Tesla is eventually able to achieve it, a humanoid form is better than single purpose robots for replacing general tasks. I also think things like portability come to mind, if you want to bring your cleaning robot and chef with you when you go on vacation and stay at a rental home, would you rather pack the 3 or more robots required or just tell your Tesla bot to get in the back seat?</p><p>I’m super interested in the topic of human robot interaction, and I think there’s a lot of work to be done in this space even with autonomous cars. Future cars should be able to communicate with the pedestrians in the crosswalk or to the other drivers around them if we want them to be truly safe. I think the question of whether or not Tesla can achieve this is a tricky one and only time will tell, but from my viewpoint, I think going humanoid is a better future than going smart house.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ece69678dab0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Augmented Reality Prototyping Tools for Head-Mounted Displays]]></title>
            <link>https://medium.com/@dylanhongtech/augmented-reality-prototyping-tools-for-head-mounted-displays-c1ee8eaa0783?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/c1ee8eaa0783</guid>
            <category><![CDATA[prototyping]]></category>
            <category><![CDATA[ar]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[hololens]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Mon, 30 Sep 2019 13:18:59 GMT</pubDate>
            <atom:updated>2019-09-30T14:17:35.561Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sKhQo8Rg7yv0DH1eU5CsrQ.png" /></figure><h3><strong>Introduction</strong></h3><p>If you recently typed “how to build prototypes for augmented reality” into the search bar, you were like me two months ago. If you were disappointed by a lack of tools, resources, or tangible example projects, you continued to be like me. I had gotten so used to having an abundance of resources from vibrant communities across the internet that could help me with anything tech related. Want to build a website? Bam! Hundreds of webinars, YouTube videos, articles, open source repositories, and more clamoring over one another to teach you the ins and outs of designing and developing a website. The same can be said for mobile applications and video games. This stark contrast in resources for augmented reality (AR) and HoloLens was quite shocking, but also very exciting. How frequently do you get to be among the first to build out functionality for an entirely new way to interact with and consume information?</p><p>The team I work on has gone through the process of designing, rapid prototyping, and testing for an AR head-mounted display (HMD) project. We thought it would be a good idea to build out some of the documentation that would have helped us along the way. This article will primarily focus on the tools we used to create our designs and prototypes.</p><h3><strong>Traditional design tools</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/907/1*Nry_n9S8jS633VzA_rRMoA.jpeg" /><figcaption>Moment from our usability testing</figcaption></figure><p>After doing some <a href="https://uxplanet.org/storyboarding-in-ux-design-b9d2e18e5fab">storyboarding </a>for the different levels, we started thinking about how we should go about building out prototypes to test with. To create most of the traditional UI elements, we used tools like Photoshop, Sketch, and Cinema 4D (I explain how we used each in the following paragraphs).</p><p>While we didn’t use this method for our project, I found a <a href="https://developer.apple.com/videos/play/wwdc2018/808/">great video</a> from Apple’s WWDC archives that show how they do low fidelity prototyping. To simulate how some information might look like in 3D space, they drew out icons and information that was then taped on clear rulers. They then opened their camera apps to see how it would feel if this design were built out. This exercise lead them to discover multiple shortcomings of their initial design such as text legibility, scaling, and information hierarchies. Even though it might feel silly, this type of paper prototyping approach is a great way to catch glaring design issues early and save time overall.</p><h3>Sketch</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*iwYKbJ2TORfVk5sLFDwkig.png" /></figure><p>Sketch is a classic design tool that is great for generating UI designs and layouts. Additionally, its export tool was great for making use of smaller components of your design like a panel with text on it or a parts of a heads-up display design. This allowed us to export all of the assets that made up a scene then use a 3D layout application to create our scenes for consumption in AR (or VR).</p><p>A good amount of our storyboarded designs were centered around floating billboards that presented information to a user, and Sketch was our tool of choice in designing those billboards. To give these UI elements the sense that they are meant to be displayed in AR, we laid out multiple assets overtop a blurred background image. This simple layout structure helped us get a better sense of how our designs would look like when they all shared the same space. Additionally, adding some tilts and warps to give the images a sense of depth and perspective also helps make these 2D screens more true to our design intentions. You can read more about our AR designs <a href="https://medium.com/@BerkShop/onboarding-design-for-ar-applications-1959d2adf333">here</a>. Using some tools later in this article, we will explore how we can take these UI elements from sketch and place them in a 3D environment relatively quickly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l4G1qPYNIKQT8APMDp1ytw.png" /><figcaption>Screenshot of our Sketch file</figcaption></figure><h3><strong>Cinema 4D</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*Z6xI01SAUMS3KVK7BbCGvg.png" /></figure><p>Cinema 4D was used to design 3D animations that might be viewed in augmented reality. By creating 3d animations, our team was able to showcase signifiers that we would be looking to build into parts of the application. These 3D assets are also great to have on hand for future use in programs like Sketchbox and Unity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*H7XtA97GMNbBSwjEcrnurw.gif" /></figure><h3><strong>Sketchbox</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*8vzYZ-pCwJwiebSycsvQag.png" /></figure><p>Sketchbox is a VR application that allows you to take 2D or 3D assets and lay them out in 3D space quickly. You do this while in a VR headset and with touch controllers, so it all feels as intuitive as placing your designs in the space around you.</p><p>Our team found that this was a great way to try to figure out how some of our more abstract design ideas might end up working out. The Google Poly integration provides a library of 3D models for you to use out of the box. This is really useful in that you don’t need to spend time creating assets because it’s more than likely that a model that is close enough to get the idea across exists.</p><p>If you have more than one VR headset in the team, there is a collaboration mode where you can load multiple people in the same design space working together. This lead to a super cool collaborative VR meeting between myself and a remote member of my team as we laid out interaction flows in the virtual room around us, talking, pointing, and drawing as we were thousands of miles away. Unfortunately, the “export to Unity” feature had some issues with packaging textures, so we were not able to use that feature in our prototyping workflow. Overall, Sketchbox 3D served as a quick way to experience and validate design concepts in 3D.</p><figure><img alt="The process of creating designs in Sketch then loading them into Sketchbox" src="https://cdn-images-1.medium.com/max/960/1*IDzA-GuaMOzgE7dWBiMcLg.gif" /><figcaption>Bringing Sketch designs into Sketchbox for VR viewing</figcaption></figure><h3><strong>Photoshop</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*U6pCsCVnulp4RExM1acKPw.png" /></figure><p>Photoshop was great for taking the AR designs from Sketchbox 3D and then contextualizing them for sharing. We found that if we took an in-app screenshot at head level, then overlaid that onto a photo taken at a similar angle, we could generate visual mockups extremely quickly (<strong>less than 30 minutes</strong>). To help with keying out the background and overlaying the AR content on the picture, I created a giant green (or blue) screen box. This helped us explain and explore different features or functionality that we had been discussing. Both to share with external stakeholders as well as to get a better sense of how implementable a concept might be.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LdcKqP8JSVzah8tekzFppQ.png" /><figcaption>Using Sketchbox to create augmented reality mockups</figcaption></figure><h3><strong>Unity and MRTK</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*LSA3lf_hGOBFEignOyVrNw.png" /></figure><p>Unity is the most complex tool that we used to create prototypes. Depending on your teams’ skillset, this might be a part of the process that you skip. However, if you’re building a HoloLens application, this will be the only way you can see your design in action in the HoloLens.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*4eWPLbpOm8iPv4_yY8zlZQ.gif" /><figcaption>Unity/MRTK to HoloLens</figcaption></figure><p>Fortunately, Microsoft has built out a large library of development resources for beginners to HoloLens development to take advantage of. I would recommend spending some time setting up Mixed Reality Tool Kit (MRTK) using resources such as this, and then looking at the prefabs in MRTK to see if anything is already built out. For example, if you are trying to test out how people interact with the HoloLens when they use it for the first time, you could just run them through the default interaction tutorials. If you’re trying to lay out buttons of different shapes and colors to see which ones people tend to like more, you will save a lot of time by using MRTK’s example buttons as a starting point and making modifications from there.</p><p><a href="https://microsoft.github.io/MixedRealityToolkit-Unity/Documentation/GettingStartedWithTheMRTK.html">Getting Started with MRTK</a></p><p>If you’re looking to bring in UI elements or 3D objects that you’ve designed for your application, you can drag them into a scene and adjust the positions and rotations to build your AR layout. From there, you can use the game viewer to navigate your scene like a video game. If you have a HoloLens on hand you can build your project to the device following a tutorial like <a href="https://youtu.be/oQvBX2GQ5zQ"><strong>this one</strong></a>.</p><p>Taking this one step further, if you are equipped to do some C# programming, you can layout different scenes and group them together in empty game objects. You can then store these different layouts in an array and use prefab buttons to cycle through them. If you wanted to make the prototypes even more realistic, you can create invisible buttons then overlay them on top of where navigation buttons exist in your designs. These buttons can use functions to activate the appropriate scenes. This would effectively allow you to build out the majority of navigating your AR designs for prototyping in a way that is reminiscent to the style of prototyping that inVision or Sketch will afford you.</p><p>If you have the time and resources on hand to use Unity as a prototyping tool, it will be able to give you the closest look at what a functional product will feel like. Just make sure you are weighing out the costs and benefits to make sure it is an appropriate tool for your team, timeline, and project.</p><p><strong>Three core steps for Unity/MRTK are:</strong></p><ol><li>Follow <a href="https://microsoft.github.io/MixedRealityToolkit-Unity/Documentation/GettingStartedWithTheMRTK.html">guides</a> to download and install MRTK, and then set up your first HoloLens development environment.</li><li>Populate Unity scenes with and 3D or 2D elements that you built out. Group assets from similar scenes under parent game objects for easy visibility toggling.</li><li><a href="https://youtu.be/oQvBX2GQ5zQ">Build out your scenes using Visual Studio</a> for deployment on the HoloLens. Now, you can view your designs in AR on the HoloLens!</li></ol><h3><strong>Powerpoint</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/100/1*xiRYQPkJqK9My3eDbUC_Xg.png" /></figure><p>When it came down the the prototyping software we actually used in our user testing, we used Powerpoint. While this might seem like an odd choice at first, Powerpoint had features that best matched what we were trying to test. One of the major features of our design was the UI layout screen. We wanted to see if we could use prerecorded videos or GIFs to explain the various bits of functionality in our UI. Neither Sketch nor inVision support embedding videos in the designs but Powerpoint does.</p><p>Using PowerPoint’s hyperlinking features, we were able to build a fully functional 2D representation of our design in just a few hours. To get around the fact that we were not delivering the information in an AR head-mounted display, we also had some users wearing the HoloLens and told them to pretend that they were seeing the screen in AR. We have an entire article about our usability testing methodology <a href="https://medium.com/@dandmendelsohn/usability-testing-for-augmented-reality-f0738ed3bb43">here</a>, but in short, the 2D prototypes greatly increased the confidence of users in the HoloLens.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*-DJUNhWa4RcSqRChupM-zw.gif" /><figcaption>Fully functional prototype using PowerPoint</figcaption></figure><h3><strong>Moving Forward</strong></h3><p>Developing augmented reality interfaces and prototypes are extremely multifaceted projects. As a team, it has been interesting to see how differently we have to think about AR design when we shift focus from handheld to Head-Mounted Displays. Additionally, interaction styles, tracking abilities, audio quality, and even user fatigue vary greatly between the few head-mounted displays that are out right now. If you wanted to learn more about our project as a whole, check out <a href="https://medium.com/@ayoraberry/prototyping-methods-for-augmented-reality-4b7687863423">this article</a> that goes over our process and frameworks.</p><p>The future hardware direction of AR is not extremely apparent at the moment, but we hope that our broad approach to prototyping and design will allow us to easily adapt to what comes next. Regardless, the best way to prepare for the future of AR design is to start designing for AR now.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c1ee8eaa0783" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Wrapping Up Thoughts on Human Machine System Design]]></title>
            <link>https://medium.com/@dylanhongtech/wrapping-up-thoughts-on-human-machine-system-design-5aa16c691307?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/5aa16c691307</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Thu, 13 Dec 2018 00:46:06 GMT</pubDate>
            <atom:updated>2019-08-27T18:25:33.761Z</atom:updated>
            <content:encoded><![CDATA[<p>We’ve covered a lot of topics over the past semester. Some that made the largest impact on me were automation, IOT, social robotics, and machine learning. I felt that (regardless of the subject we were covering) the best discussions always surrounded the worries around privacy and ethics.</p><p>I absolutely love consumer technology. It probably consists of over 10 percent of all things that I think and talk about. I’ll even be sprinkling this post with some photos of my Silicon Valley vacation travels. However, as these human-machine systems are pushing the boundaries of what products can do, they are also pushing the boundaries in terms of how much information they need to know about us. Companies stake their livelihood on the success of automated systems. Many times, these systems use machine learning which makes these “black boxes”. The algorithms make their own decisions and can’t explain them to the people who developed them. If companies are willing to trust the machine-learned process, will we?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*qx02V_z1IiI-K46cQuL4HQ.jpeg" /><figcaption>Picture of me in Google’s Android garden</figcaption></figure><p>Will people stop holding onto the luxury of choice if robots can just make better decisions for us? Will people stop manually giving smartphone applications access to microphone, location, or camera data if sharing everything becomes standard?</p><p>I ask these questions because I really have no idea what people will be willing to do. We’ve clearly allowed Amazon to place always-on microphones and even cameras in our homes, and we allow Google and Apple to track our physical locations. Will there ever be a limit? When I think about what people fear, I am reminded that sometimes people fear what they don’t understand. Sure, sometimes public outrage is warranted, but sometimes massive groups of people will protest something that only exists to make their lives more livable (think anti-vaxers). What if public fear of machine learning halts the progress of a technology that could make our lives significantly less stressful?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*CQWq12RWu_JnObOgJ027xw.jpeg" /><figcaption>Picture of me at Apple’s (now old) campus in California!</figcaption></figure><p>When looking at videos of Twitter, Facebook, and Google CEOs speaking before congress, it becomes painfully clear that leaders of this country have no understanding of these massive social platforms.</p><p><a href="https://www.wired.com/story/congress-sundar-pichai-google-ceo-hearing/">Congress Blew Its Hearing With Google CEO Sundar Pichai</a></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FT_Jt-0b6M1w%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DT_Jt-0b6M1w&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FT_Jt-0b6M1w%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/368057dc54ab523731d6ba9fb7a22e82/href">https://medium.com/media/368057dc54ab523731d6ba9fb7a22e82/href</a></iframe><p>I realize that for a “Wrapping up” post, I’m asking way more questions than I’ve answered, and that’s pretty exciting. I am optimistically fearful of the technology that is on the horizon. I don’t even know where I stand on the privacy/convenience debate. Although, I tend to be an early adopter of technology, and if Elon Musk needed to know my shower and sleep schedule to get self driving cars on the road quicker, I would’t hesitate to hand my data over.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*AMSCYF4PhdO4K1FnO45ueA.jpeg" /><figcaption>Edgy high schooler makes a joke about Facebook’s “like”</figcaption></figure><p>During this past semester, we’ve asked questions about the future, developed automated systems, and analyzed trends in technology. Building your own system opens up your mind to the possibilities of (invasive but awesome) IOT projects using your location, your schedule, and your emotions. We could tailor every day to how we feel. We can tailor every customer service interaction if there is no human on the other end and a conversational AI knows everything about you. We could give all of our problems to a cloud-based system and take a nap while the algorithm solves them for us. But is this lack of stress good for us? What will we do?</p><p>Every time I started trying to share something I’ve thought about this semester, I’ve asked more questions. I’m excited and scared about the future of technology, but I’m glad to have had the opportunity to discuss these all in an academic setting. I hope to explore this role of technology in our lives professionally soon. If future me can find a balance between control, autonomy, and convenience, I’d be pretty darn impressed.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5aa16c691307" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Chatbots]]></title>
            <link>https://medium.com/@dylanhongtech/chatbots-190704db118e?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/190704db118e</guid>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Sat, 08 Dec 2018 23:01:47 GMT</pubDate>
            <atom:updated>2018-12-08T23:01:47.521Z</atom:updated>
            <content:encoded><![CDATA[<p>I remember when Siri on the iPhone 4s was first announced. I went to the Apple Store just to try out the feature because I really couldn’t believe that I could ask my phone (with my voice!) to do something and it would actually try to do it for me. Now, this is pretty common. Now, people have Echos and Google assistants in their homes and the voice recognition and logical processing has gotten good enough where I’m actually willing to use it. But for now, we’re all very aware that there isn’t another person on the other end. We know that we’re talking to an algorithm. When will we lose the right to that knowledge?</p><p>When looking at Google Duplex, a voice assistant that actually acts like your assistant, it’s clear that we are approaching that lack of transparency. Even now, customer service chat bots that we type to interact with are not always human.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FD5VN56jQMWM%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DD5VN56jQMWM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FD5VN56jQMWM%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/139d9aa3e58d35d8e9a3e6c46a169b71/href">https://medium.com/media/139d9aa3e58d35d8e9a3e6c46a169b71/href</a></iframe><p>I was recently watching a video that compared many different call center experiences when trying to purchase a prebuilt computer over the phone. I’ll link to the twenty minute video below, but the main point was that the process generally sucked. Also, there was a lack of any type of consistency between the different callers. The people were very different! They had varying levels of knowledge about computer parts, had separate methods of trying to sell things to people, and sometimes they really didn’t seem to know much about what they were selling. If you are calling a store with the purpose of buying a PC, you probably want to know what your getting yourself into.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FrzpYkpZX8qw%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DrzpYkpZX8qw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FrzpYkpZX8qw%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/efdae8b13ee18db9368cb4917f11bc80/href">https://medium.com/media/efdae8b13ee18db9368cb4917f11bc80/href</a></iframe><p>Chatbots could theoretically alleviate these issues. They could be completely fluent in whatever field they are being used for, and customers could always have a consistent experience.</p><p>There are some places where I do think that talking to a robot is better than a person. I already make that choice when I order online as opposed to calling a business because I usually find the online process more streamlined. And if vocal chatbots allow for a more streamlined process while my hands are busy… I guess I’d be good with that too.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=190704db118e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[IOT Relfections]]></title>
            <link>https://medium.com/@dylanhongtech/iot-relfections-4e31d4686090?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e31d4686090</guid>
            <category><![CDATA[internet-of-things]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Fri, 30 Nov 2018 17:50:54 GMT</pubDate>
            <atom:updated>2018-11-30T17:50:54.827Z</atom:updated>
            <content:encoded><![CDATA[<p>As we begin wrapping up the semester, this is a good opportunity to think about the core technology that enables most of modern automation. When we think about social robotics, wearable devices, and machine learning systems, most of these require a cloud-based network to share information across a intelligent system. The Internet of Things is this backbone. Smart-connected devices allows us to make “dumb” products… smart! For wearable technology, most of the sensor data is useless if it’s not analyzed for trends, compared against recommended baselines, or stored in an easily viewable format.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*H-yN6ffN1mpSvuMuK4QhWg.jpeg" /></figure><p>IOT is a great way to piece together different innovations across consumer, medical, and industry technology. However, developing these systems still require a fair amount of user input. The “smart home” experience is far from seamless. The closest implementation I’ve seen to an easy setup is Apple’s HomePod, but that exists within the “walled garden” of Apple’s ecosystem. In a more popular setup (Google Home or Amazon Echo) there are third party hardware and software solutions that run on completely different infrastructures. For example, the Philips Hue smart bulbs require a home base to communicate with all of the products, but my smart switches in my room communicate through a server managed by the hardware developers. It’s really quite a mess.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fWrdvSSUTcMqa2iGqaKulA.jpeg" /></figure><p>As my final project team has begun to see, IOT starter kits such as the Wio Link are quick ways of getting started developing your own IOT experiences, but they still require a fair amount of development if you do not want to be constrained to the prebuilt modules that other companies have created. IFTTT is a fantastic service, and it’s doing a great job of creating a library of these applets, but it’s still hard to take the backbone of an existing applet and fine-tune it to exactly what you want it to do (if that functionality does not already exist).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*IzjnZN1CeBdnEl0I0yiLqQ.jpeg" /></figure><p>I think IOT has been around us for longer than we realized, but it’s just now that it is becoming modular enough for (more) people to experiment with it on their own and potentially create IOT solutions for themselves. I think this accessibility will bring in a huge new wave of problem solvers in k12 education spaces. Also, this accessibility really gets people thinking about how devices can network together. And prototyping has become so much easier because most of the easy-to-purchase starter kits negate the need to design custom boards and chips! IOT will find its way into more ordinary devices, finding patterns we never thought to look for, and I’m excited to see what people find.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e31d4686090" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Navigation and GPS Through Augmented Reality]]></title>
            <link>https://medium.com/@dylanhongtech/navigation-and-gps-through-augmented-reality-cd8012dfc8f2?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd8012dfc8f2</guid>
            <category><![CDATA[augmented-reality]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Fri, 09 Nov 2018 18:23:04 GMT</pubDate>
            <atom:updated>2018-11-09T18:23:04.156Z</atom:updated>
            <content:encoded><![CDATA[<p>GPS navigation has become a standard for getting people from one place to another. Physical paper maps are pretty much ancient when it comes to navigational technology. I remember when I was in middle school and my parents purchased a TomTom GPS system for our car. Even as a kid, it was clear how important this technology would be moving forward. I could tell that planning a long trip using MapQuest was a pain and that it was something that I wouldn’t have to do by the time I learned how to drive. Now, people don’t even need to purchase a separate product to reap the benefits of GPS. With Google maps providing turn-by-turn directions to all smartphones, this extremely powerful and convenient tool is available to everybody.</p><p>Navigation is still not seamless. Frequently, people will have their phone mounted to their dashboard or laying flat on their center console. This means that people need to look away from the road to get the information they want. Not only is this dangerous, but it’s also a clunky solution. With augmented reality (AR) technology, mapping information will be overlaid onto the real world, allowing us to focus on things like driving and walking. Check out Google’s AR extention to Google Maps.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F4F0gFpzsYLM%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D4F0gFpzsYLM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F4F0gFpzsYLM%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/3ff8b8dab76f9a94686c55a51aa4b0c2/href">https://medium.com/media/3ff8b8dab76f9a94686c55a51aa4b0c2/href</a></iframe><p>Phone AR is still not seamless. I see modern AR as a stepping block toward the future of information. When eyeglasses become technology wearables, this type of data overlay will become much more seamless. Google Glass had the right idea conceptually but were never able to deliver on their promises on the hardware side of things. And if you doubt that people will make the change to wearing these glasses, think about how weird it is that people started carrying around phones in the first place. People who didn’t wear watches started wearing fitness trackers and smartwatches because they added value into daily life.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FIZdkIVS53Uw%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DIZdkIVS53Uw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FIZdkIVS53Uw%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8852ecaf4aa1a03e5e05a34b8bc701f4/href">https://medium.com/media/8852ecaf4aa1a03e5e05a34b8bc701f4/href</a></iframe><p>Heads up displays (HUDs) will be able to overlay navigational information onto the roads using the windshield as a medium. However, after scraping the web, current HUDs look like they do either the same or less when compared to smartphone navigation. They simply project a screen onto the windshield as opposed to overlaying information onto the surrounding environment. I am doubtful that HUDs will become compelling enough before self-driving cars remove the need for driving navigation tools completely. Although, AR through HUDs could be a cool method of displaying what the car is going to do to passengers.</p><p><a href="https://www.digitaltrends.com/cars/best-head-up-displays/">Head-up displays bring fighter-jet tech to your car in the name of safety</a></p><p>GPS data is so important to visualize and the current suite of mapping software has done a really good job of conveying location data. However, I see AR being the next major stepping stone of GPS technology and devices. AR navigation through a headset or eyewear gets so close to the sci-fi dream of having a chip implanted into your head. If you need to know how to get somewhere, the information is seamlessly relayed to you in a way that you don’t even have to think about. You simply know how to get anywhere! And with Apple working on mapping inside spaces such as malls and airports, the future of navigation will finally live up to those Google Glass commercials from many years ago.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd8012dfc8f2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Emotive Display and Social Robotics: My Opinions]]></title>
            <link>https://medium.com/@dylanhongtech/emotive-display-and-social-robotics-my-opinions-d76c459c42b1?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/d76c459c42b1</guid>
            <category><![CDATA[robotics]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Mon, 05 Nov 2018 00:58:08 GMT</pubDate>
            <atom:updated>2018-11-05T00:58:08.444Z</atom:updated>
            <content:encoded><![CDATA[<p>As opposed to a more research heavy post, I’m going to focus on some of my personal opinions with regards to emotive and social robots. I have yet to see one social robot that has been compelling enough to seem like a finished and useful product. The initial Jibo indiegogo promotional video used to show off the supposed future of home robots greatly overestimated how well the technology would link up with people’s lives.</p><p><a href="https://www.indiegogo.com/project/jibo-the-world-s-first-social-robot-for-the-home/embedded">Indiegogo: JIBO, The World&#39;s First Social Robot for the Home</a></p><p>I’m not against the idea of social robotics. I think they will have plenty of uses eventually. The future in which teachers, therapists, and caregivers are more accessible is a bright one. However, all of these human-occupied roles are multidimensional. The ability to go off on a tangent and elaborate on something not in the textbook makes somebody a good teacher, but many contemporary social read from a script.</p><p>I feel that the personal assistant/artificial intelligence component of social robots just isn’t there yet. If Amazon, Google, and Apple (some of the worlds’ most valuable and wealthy companies) can’t make a personal assistant that feels as natural to use as talking to a person, how is a small startup going to do that while also building a physical/personable shell <em>and </em>trying to<em> </em>create a stable emotional connection with the user? Yes, Google Duplex is pretty cool, but they have one primary goal and it’s in a conversational field that is as transactional as they get.</p><iframe src="https://cdn.embedly.com/widgets/media.html?url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dbd1mEm2Fy08&amp;src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fbd1mEm2Fy08&amp;type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/b135a07c2e9d691397872e99737375a7/href">https://medium.com/media/b135a07c2e9d691397872e99737375a7/href</a></iframe><p>What social robotics needs is time. Which means that the companies that are trying to accomplish these tasks should keep going. It’s just that the marketing for their products are getting out of hand. It’s possible to kill a mediocre product with incredible marketing and I’d hate to have another Google Glass situation on our hands in a field that’s largely focused on social good. Maybe once Siri stops responding to “add half and half to my shopping list” with “One half plus one half equals one, Dylan.” social robots might have a chance.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d76c459c42b1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When Will We Catch Up to Big Data?]]></title>
            <link>https://medium.com/@dylanhongtech/when-will-we-catch-up-to-big-data-e9c7982c8ab0?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e9c7982c8ab0</guid>
            <category><![CDATA[big-data]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Fri, 19 Oct 2018 17:17:46 GMT</pubDate>
            <atom:updated>2018-10-19T17:23:38.777Z</atom:updated>
            <content:encoded><![CDATA[<p>It’s easy to collect information. So much of our personal data is stored on our phones then sent out to countless servers that all sort and collect different bits and pieces of who we are (and we probably opted in to this process at some point). The companies that wanted your information really didn’t have to work that hard to get it. You and I willingly gave it up because we wanted the benefits of some service.</p><p>So there’s this idea that companies are going to know us better than we do. Shipping products we need before we think we need them or scheduling doctors appointments before we feel sick are all within the realm of possibility when it comes to autonomous decisions based on big data. But something that we don’t always think about is how difficult managing big data can be.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/414/1*nZ40rxf6GhsRuz5RqOwNSQ.gif" /><figcaption>Tesla’s Robotic Factory</figcaption></figure><p>I was recently talking about data visualization in Augmented Reality (AR) for a factory system. Remember how I said that collecting people’s information is easy? It has also become extremely cheap to collect position, environment, and efficiency information about machines. All actions must be computed through a chipset, so why not store those somewhere? And all the sensor information that is required to make a robot work, store that information somewhere too. So now, we have these massive factories sending floods of information to servers (because why not?) but there is no system in place to create any value from it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/989/1*mBqZczk8tRFum-oV7vDu2g.jpeg" /></figure><p>Just because your information is being stored somewhere doesn’t mean it’s being put to any sort of use. Companies have been storing as much as they can without seeming too big brother-ish (sometimes) in the hopes that they will be able to do something with it later. Even the subscription service MoviePass was built on the hope that data would be worth enough to offset the price of paying for everybody’s movies. If you haven’t been following the MoviePass saga, I’d highly recommend looking into it. It’s a great example of data collection without a clear vision.</p><p><a href="https://www.theverge.com/2018/10/3/17933582/moviepass-subscription-ted-farnsworth-investment">MoviePass subscribers aren&#39;t going to the movies, and that&#39;s what MoviePass wants</a></p><p>Big data is pretty crazy because it’s too much information for humans to really understand. But we’ve been smart enough to develop tools to help ourselves visualize and act on too much information. Machine learning is likely going to be a key component in figuring out how make an image out of all the noise we’re collecting. However, in it’s current state, sometimes we are collecting more information than we can handle. Hopefully we figure out what to do with it before the hard-drives our information is stored on become obsolete.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e9c7982c8ab0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Designing The Future Human]]></title>
            <link>https://medium.com/@dylanhongtech/designing-the-future-human-f12f4e9f5904?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/f12f4e9f5904</guid>
            <category><![CDATA[future-human]]></category>
            <category><![CDATA[medical-technology]]></category>
            <category><![CDATA[apple-watch]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Fri, 12 Oct 2018 21:45:13 GMT</pubDate>
            <atom:updated>2018-10-12T21:45:13.705Z</atom:updated>
            <content:encoded><![CDATA[<p>I recently picked up the Apple Watch Series 4. It’s really incredible how powerful of a computer it is. Check out some articles that show the watch computing physics based Metal rendering while still having CPU headroom for more calculations.</p><p><a href="https://bgr.com/2018/10/03/apple-watch-series-4-processor-speed-and-performance-explained/">Apple Watch Series 4 might be more powerful than the iPhone in your pocket</a></p><p>This was a massive upgrade for me, as four years ago I purchased the Series 0 Apple Watch on day 1. When that watch launched, it was unable to load third-party applications on its own without offloading all of the processing to its parent iPhone’s chipset. In four years, the increase in computational ability for a computer that we wear on our wrists has seen enormous improvements. Using this line of thought, there will soon be computers that we put on other parts of our body. Augmented reality glasses, smart contact-lenses, or even brain implants could be a part of an everyday future human.</p><p>One of the reasons people love the Apple Watch are its health benefits. I’ve linked to a pretty cool promotion that Apple’s marketing team put together using letters that people wrote to them about the watch. This, with the help of some life saving heart sensors, has become a piece of technology that some people rely on to stay healthy.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FN-x8Ik9G5Dg%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DN-x8Ik9G5Dg&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FN-x8Ik9G5Dg%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/05974bd781b3b6ef2c8fcf0ca05e6fe6/href">https://medium.com/media/05974bd781b3b6ef2c8fcf0ca05e6fe6/href</a></iframe><p>I would bet that future technology based health accessories will do more than monitor and diagnose. In most cases, I expect that smart devices will also be able to administer treatment in the future. Perhaps even providing autonomous treatment for illnesses we haven’t even begun to feel the symptoms for. Even the process of getting smart-implants could be automated through the use of robotic surgeons.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/260/1*mDOXk-TvORSLcdpMlhChqA.gif" /><figcaption>Hopefully the robotic surgeon is better than this guy</figcaption></figure><p>Humans historically develop technology to create convenient solutions to annoying problems. Now that we’ve (almost) figured out air and land travel, now that we can send information across the world in a matter of milliseconds, and now that we’ve figured out how to use a voice assistant to order delivery from the nearest takeout place so we never have to have conversations with actual people, we will probably figure out how we can live our lives without pain, illness, or any of the physical struggles that traditionally come with human life. In the same way that we think back on people who lived life without cars or the internet as ancient and monochromatic, maybe the future humans will receive school lessons implanted into their brains about how previous iterations of humans used to get “seasonal allergies”.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f12f4e9f5904" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using Automated Signal Detection to Police YouTube Content]]></title>
            <link>https://medium.com/@dylanhongtech/using-automated-signal-detection-to-police-youtube-content-ab04cfd1a396?source=rss-457f74ce468d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab04cfd1a396</guid>
            <category><![CDATA[demonetization]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[youtube]]></category>
            <dc:creator><![CDATA[Dylan Hong]]></dc:creator>
            <pubDate>Thu, 27 Sep 2018 21:02:31 GMT</pubDate>
            <atom:updated>2018-09-27T21:02:31.717Z</atom:updated>
            <content:encoded><![CDATA[<p>Using artificial intelligence or machine learning to flag and remove content online has been a growing trend among top social media sites. Famously, Facebook has been hiring extra staff to manually sift through posted content to reduce “Fake News” because its automated systems haven’t been performing as well as the world would like.</p><p>My personal experience with automated signal detection revolves around YouTube content. I’m the type of person who watches way more YouTube than TV. I actually produce YouTube content from time to time (but we’ll get back to that later). If you’ve been following any particular YouTuber over the past two years, you’ve probably heard something about rampant demonetization. Advertisers have been getting upset at YouTube because their adds have been showing up on racist, offensive, or inappropriate content. Despite YouTube’s development for tools to help advertisers control where their adverts show up, large corporations have been pulling their commercials from the platform.</p><p>YouTube’s response has been to use machine learning algorithms to automatically flag and demonetize (remove ads) uploaded content that doesn&#39;t follow the community standards and guidelines. Things like hate-speech, violence, and sexual content are meant to be automatically flagged. However, because there is no human oversight, massive creators on the platform have been losing revenue due to their videos not displaying advertisements.</p><p><a href="https://youtube-creators.googleblog.com/2018/04/an-update-on-our-2018-priorities.html">An Update on Our 2018 Priorities</a></p><p>In this official statement from YouTube, they allude to the idea that they are still trying to update their systems that auto-flag videos. While this process occurs, the accuracy seems to be increasing only be incrementally.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1004/1*AJqeDZRxfMgaOCAgFfYGxg.png" /><figcaption>MKBHD’s tweet about video demonetization</figcaption></figure><p>One of the most notable demonetization storms happened with the release of the iPhone X. About a year ago, much of the tech community on YouTube began posting their thoughts on the latest and greatest iPhone; however, nearly all videos that included iPhone X in the description or title were demonetized for some period of time. The theory is that the algorithm mistook “iPhone X” to somehow be related to the term “XXX” which would, in fact, be demonetize-worthy content. iPhone reviews are massive in the technology YouTuber space, so many large reviewers put extra production time into these videos. Having them demonetized for the first 24 to 48 hours could lead to negative profits for the production of a particular project.</p><p><a href="https://www.theverge.com/2017/11/14/16648348/youtube-demonetizing-iphone-x-videos">YouTube creators are frustrated that a bot keeps demonetizing their videos</a></p><p>This <em>The Verge </em>article does a great job covering the incident. The article points out that “ [c]reators just want YouTube to handle communication differently. ‘I trust that YouTube is working hard on their end,’ Hong said. ‘What YouTube is trying to do from a machine learning standpoint is really difficult… but their community engagement side is terrible.’” And the man has a point! A machine learning system at this scale that is trying to understand videos and audio to try to determine if it contains inappropriate content is a wildly ambitious goal. However, it should be on YouTube to maintain their human input while the automated, signal processing, system is not quite up to snuff.</p><p>Eventually, machine learning and artificial intelligence will become the backbone of all signal processing. From TSA scanners to finally being able to detect that I haven’t taken my bag of whole grain Tostito’s from the bagging area, humans will be less involved with these types of redundant tasks. For now, companies that are rolling out this reliance on AI for signal processing need to be transparent and quick to intervene when things go poorly.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab04cfd1a396" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>