<?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 Anna Zawilska on Medium]]></title>
        <description><![CDATA[Stories by Anna Zawilska on Medium]]></description>
        <link>https://medium.com/@anna.zawilska?source=rss-eea3738ca108------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*t56wJNX9EDlLkevP.</url>
            <title>Stories by Anna Zawilska on Medium</title>
            <link>https://medium.com/@anna.zawilska?source=rss-eea3738ca108------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:47:11 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@anna.zawilska/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[Humans aren’t rational agents — and that’s why you need User Research]]></title>
            <link>https://medium.com/@anna.zawilska/humans-arent-rational-agents-and-that-s-why-you-need-user-research-d73176a1a316?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/d73176a1a316</guid>
            <category><![CDATA[ux]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Tue, 06 Mar 2018 14:24:09 GMT</pubDate>
            <atom:updated>2018-03-06T14:24:09.006Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i1TzjLxqhkGaNWLFWGPRcQ.jpeg" /><figcaption>Image from <a href="https://www.pexels.com/photo/white-black-tower-chess-69128/">Pixabay</a></figcaption></figure><blockquote><em>Why don’t we exercise more, though we know we should?</em></blockquote><blockquote><em>Why is it difficult to save for retirement?</em></blockquote><blockquote><em>In the case of a scary health diagnosis, why does it matter for a doctor to tell us in-person?</em></blockquote><p>To be at times <a href="https://en.wikipedia.org/wiki/Rational_agent">irrational</a>, is to be human. Under the influence of evolutionary and societal forces, we can act in ways that seem sub-optimal with reference to important goals.</p><p>Understanding these ‘irrationalities’ is critically important to product development. It means the problem space might be more complicated than expected. <a href="https://techcrunch.com/2017/10/28/why-snapchat-spectacles-failed/">Naive solutions that don’t take into account how people act, think, and feel in practice are doomed</a>.</p><p>And that’s why User Research matters in the earliest stages of a startup’s life. It’s a tool to interrogate the problem space, and understand any irrationalities in human behaviour within it. This serves as a critical foundation for formulating the purpose and features of your product.</p><p>In healthcare, there are plenty of scenarios where people’s psychology and behaviours might be surprising. At babylon, we have to take these into account when designing products to help people achieve good health. For example, we don’t always act in the interest of long-term goals even if we know what to do. We will eat badly or avoid exercise, even though we know the consequences. Just telling a user they need to do something, won’t necessarily drive behaviour change towards good health. As another example, the same piece of health information can feel very different coming from the mouth of a doctor than an alternative information delivery channel like a written report. We have to think deeply about the value of a human doctor in healthcare, and not assume all information can or should be delivered through an app.</p><p>As a researcher or product manager, you don’t have to start from scratch in understanding user psychology. There are theories and concepts in Behavioural Psychology (and surrounding fields), that can help structure research inquiry and move you faster to critical insights. As a starting point, I suggest researchers and product managers draw inspiration from <a href="https://www.thisisproductmanagement.com/episodes/motivating-users/">self-determination theory</a> and <a href="https://www.thisisproductmanagement.com/episodes/behavioral-psychology/">competing pressures</a> model of behaviour.</p><p>Babylon are one of the few startups at the forefront of understanding and unpacking human psychology when it comes to healthtech. If this excites you, <a href="https://www.babylonhealth.com/vacancies/">join us</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d73176a1a316" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Challenges of Designing an App to Communicate Health Information]]></title>
            <link>https://medium.com/@anna.zawilska/the-challenges-of-designing-an-app-to-communicate-health-information-78fcc4b6afba?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/78fcc4b6afba</guid>
            <category><![CDATA[health]]></category>
            <category><![CDATA[ux]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Mon, 05 Feb 2018 14:49:19 GMT</pubDate>
            <atom:updated>2018-02-05T14:49:19.357Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vx_khh8xXublRZiTzD4CCw.png" /></figure><p>Doctors spend their whole careers honing how they communicate health information (e.g. disease risk, test results, recommended actions) to their patients. These communication skills are critical to providing a high quality healthcare experience. (Studies have shown that doctor-patient communication breakdowns are a dominant cause of malpractice claims*.)</p><p>In order to achieve our mission of making healthcare scalable and accessible, we have to design for the scenario that our app serves in place of a human doctor in some appropriate situations. And in order to deliver an excellent healthcare experience, we need to create an app interface that communicates health information well, kind of how a doctor might.</p><p>This is a big challenge as there are no well-established design guidelines or best practice for this. Here are some of the key issues we are tackling at babylon as we design our app to deliver health information well:</p><ul><li>Resolving questions</li></ul><p>In many scenarios, patients have questions and doubts in response to health information. In human-human interaction, it’s relatively easy to detect whether the other party has understood and accepts a piece of information by seeing their body language and how they speak. In a traditional consultation, a doctor can detect this and proactively try to repair the issue, iteratively, until the patient appears satisfied. It is extremely difficult to give an app these dynamic capabilities in an app-patient interaction.</p><p>At babylon, we do a lot of rigorous research before and after launches to maintain and develop a taxonomy of questions and concerns users might have around different types of health information. We then integrate the right kinds of information at the right time into the user journey to proactively reassure and resolve them.</p><ul><li>Delivering distressing information with sensitivity</li></ul><p>Human doctors are usually highly tuned to the individual patient’s personality and mood, and the nature of the message they are trying to deliver. They can deliver it in the right way. A distressing message on health should be communicated totally differently to a positive one.</p><p>In our interface, we know multiple design guidelines around presentation of information are needed depending on the user, user journey, and nature of information to-be-communicated. We have been learning the best ways to communicate information without creating unnecessarily alarm by working closely with our users and our in-house doctors. This is especially important as we offer an improved user experience over Googling symptoms, which often leads to unnecessarily distressing outcomes.</p><ul><li>Establishing credibility</li></ul><p>Information delivered by a human doctor is, typically, automatically more credible in the eyes of users than the same information in an app. Internally at babylon, we believe completely in the rigour with which we build our products and can trust in the information delivered. All our products are heavily supervised by doctors and regulated by medical authorities. However, that doesn’t automatically mean users buy into the credibility of our product and the information it offers.</p><p>We are continuously learning about the best ways to build trust in our product. For example: Is the solution to expose how our system works to the end user? Given the complexity of the machine learning model that powers our app, what’s the best way to do this? Is the solution exposing the deep involvement of doctors and medical regulators in all we do?</p><p>These themes just skim the surface of our ongoing work unpacking and understanding how to create an app that provides an excellent healthcare experience. More Medium articles on this to come, so watch this space.</p><p>It’s extremely challenging but rewarding to be working at the forefront of developing design guidelines and best practice in health-tech. If this kind of research and design work is meaningful to you, get in touch or apply for <a href="https://www.babylonhealth.com/vacancies/">one of our open roles</a>! We’d love to hear from you.</p><p>Thanks so much to my stellar colleagues <a href="https://medium.com/u/b43b5971548">Adrien Lassalmonie</a> and <a href="https://medium.com/u/4e08f2909690">Adrisketch</a> for helping me edit and evolve this post! You rock.</p><p>*e.g. <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1201002/">https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1201002/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=78fcc4b6afba" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using field research at startups to move faster and reduce risk]]></title>
            <link>https://medium.com/@anna.zawilska/using-field-research-at-startups-to-move-faster-and-reduce-risk-39fc7352b7b0?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/39fc7352b7b0</guid>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[healthcare]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Tue, 10 Oct 2017 11:03:26 GMT</pubDate>
            <atom:updated>2017-10-10T11:03:26.082Z</atom:updated>
            <content:encoded><![CDATA[<p>Understanding healthcare in practice today, to shape healthcare for the future</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/940/1*fYGg40DtGkolikxa2HVp0A.jpeg" /></figure><p>At babylon, our eyes are firmly focused on a future where technology and Machine Learning open up affordable and accessible healthcare to everyone on earth. And yet, while we and other disruptive startups set our sights on the future, it’s important we continue to ground our work in an understanding of the present. In order to deliver a better healthcare service, we need to be aware not only of the key issues to-solve, but also what parts of the current user experience are valuable and delightful (and why) so we can bring this forward into our product.</p><p>Our team employs a variety of research methods including usability study, survey, and interviews, but there are certain things about current healthcare that our participants will never think to tell us because they are so accustomed to navigating the current healthcare system; the only one they’ve ever known. For example, for doctors in particular, there’s a whole set of tacit communication and interaction strategies they employ, honed over time, to deliver high quality healthcare from both clinical and patient experience perspectives. There’s also a large rift between theory and practice when it comes to physical spaces of healthcare delivery. For example, leaflets are often set up in waiting rooms with the intention to inform and empower patients. However through observation it becomes quickly obvious how little patients look at those leaflets. Field research — spending time in traditional GP Practices — helps shine a light on aspects of healthcare that are otherwise not visible.</p><p>When doing fieldwork, it’s critical to proactively try take the perspective of an outsider: someone seeing the setting and the activity within it for the first time. As a researcher, you should try to make the familiar or mundane appear strange, questioning why things are the way they are.</p><p>Even just a few hours of field research can generate important insights for your business:</p><p><strong>1) Metrics for user experience</strong></p><p>There is a lot to be learnt about metrics by looking at the traditional healthcare system. Take a doctor-patient interaction, for example. It turns out that doctors implicitly understand that making sure patients feel heard is critical to delivering both good quality clinical care but also good patient experience. For example, doctors typically give patients a critical few minutes to talk freely and openly both about their symptoms but also anything else that’s concerning or worrying them. Then, later in the consultation, doctors ensure they mention and address all the points the patient mentioned. It turns out this is a critical part of delivering an excellent patient experience. In the world of remote consultations with doctors, or when recreating a doctor consultation through a chatbot, rather than simply asking for a generic NPS, we can ask users about how understood and heard they felt.</p><p>Using domain-specific metrics will probe more meaningfully and directly at a critical part of user experience in healthcare. These metrics are very actionable because they recommend a more particular facet of the user-product interaction that we need to evolve, than a generic NPS score. From this, we can kickstart solution generation and validation. For example, if we find users of our symptom checker chatbot don’t feel understood or heard, we can develop and test a set of UI components to tackle that. We might develop new ways for users to provide input, or we might add a UI component that better explains and ties the chatbot’s response to the user’s input.</p><p>By using the metrics particular to the domain, we save ourselves the time and effort uncovering the root interactional issues. We avoid re-inventing the wheel.</p><p><strong>2) Pre-existing models in users’ minds</strong></p><p>Field research helps us see in practice the steps and ideas people have to employ to successfully complete the tasks they want to in the traditional healthcare system. This helps us build up an image of the mental model that users have around healthcare, and gives us a clear view on the gap that exists between the workflows and concepts they’re used to, and the ones in the new version of healthcare we are creating and shaping.</p><p>This understanding is critical in designing onboarding and in crafting marketing messaging. As a disruptive startup, the onus is on us to take the users on the journey from the healthcare they know, to a new version that delivers even more value and delight. If we know what users are used to, then we know what might surprise them and what might delight them. We can also pre-empt any major user concerns or uncertainties and deal with them proactively in messaging within marketing and the product.</p><p><strong>3) Focusing on opportunities for 10x improvement</strong></p><p>Understanding the current patient journey, helps you to see the opportunities to create a new offering that is 10x better than the current system. This 10x improvement, rather than an incremental one, is critical to increasing the success rate of getting users to change from the current system to your one. As we know, people are very resistant to change, so your offering needs to be a substantial improvement.</p><p>Often, some of the pains of the current system will be obvious to you. For example, at babylon we know that sometimes people struggle to get appointments with NHS GPs without a long wait. So, we’re focused on solving that problem. However, other pains might be less obvious, requiring a closer look at the patient experience. For example, if you follow the day-to-day work of a GP Practice, you’ll see how much physical and emotional harm, not to mention cost to the NHS, comes from the fact that our healthcare is very reactive rather than proactive. By the time a patient is seen and treated, even small cheap easy health issues may become big expensive difficult problems. At babylon, we’re asking ourselves how we can help people overcome this, and empower them to be more proactive. Research helps us quantify the size of the opportunity and tangibly understand impact it could have on the patient experience.</p><p>Fieldwork can sometimes be perceived as time-consuming for little reward. However, on the contrary, for product teams aiming to revolutionise and transform an existing service, having a solid understanding of the current system will help you define and measure success more effectively. Ultimately, this will save teams time and resources, helping them move faster with minimised risk towards a product that truly delights and delivers value. With even just a few hours spent in the field, you can check your blind spots and avoid spending resource on achieving only a local maxima in user experience.</p><p>If you want to play your part in defining the future of healthcare, we’d love to have you on our team! Please take a look at our <a href="https://www.babylonhealth.com/careers">current vacancies</a> and don’t hesitate to reach out.</p><p>Major thanks to my colleagues <a href="https://medium.com/u/b43b5971548">Adrien Lassalmonie</a> and <a href="https://medium.com/u/4e08f2909690">Adrisketch</a> who read my article and gave me feedback!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=39fc7352b7b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Learning by doing: my first few weeks as a User Researcher at babylon]]></title>
            <link>https://medium.com/@anna.zawilska/learning-by-doing-my-first-few-weeks-as-a-user-researcher-at-babylon-b26bce562bb9?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/b26bce562bb9</guid>
            <category><![CDATA[user-research]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[health-technology]]></category>
            <category><![CDATA[babylon-health]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Tue, 15 Aug 2017 16:52:56 GMT</pubDate>
            <atom:updated>2017-08-21T16:07:33.833Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2ec4452PiJNymYejKIh8NA.png" /></figure><p>I’ve recently finished month one as the first full-time User Researcher at babylon; it’s been fun, challenging, and I’ve learnt a heck of a lot. It’s extremely rewarding and motivating to be working towards babylon’s mission that sits at the heart of all we do: to bring accessible and affordable healthcare to everyone on earth. Reflecting on my first month, I see many learnings around settling in as a User Researcher at a startup. I’m sure there will be many more! Here are some of the key things I’ve learnt so far:</p><h3><strong>Don’t wait to be given opportunities for research; probe and help define them.</strong></h3><p>The first week of your new User Researcher role should be all about gaining an understanding of the different product areas, the organisational structure, and beginning to develop a picture of others’ challenges which you could alleviate with research.</p><p>In your first days, work with your line manager to generate a list of product owners and tech leads to meet. Then spend your first week interviewing them to learn about their products, the stories behind what they’re working on, and their biggest risks and unknowns. Use your research skills here to begin to build a deeper understanding of your team and their needs. I recommend <a href="https://library.gv.com/questions-to-ask-before-starting-user-research-4607c2633f6f">using these questions by Michael Margolis</a> as a basis for your interviews.</p><p>In addition to learning about product, aim to assess how comfortable and familiar people are with user research, which will help you craft a collaborative approach. Some team members may have less experience, in which case proactively involve them throughout your research to show them how it’s done and address any uncertainties or concerns.</p><p>In the fast-paced startup setting, the key user research opportunities won’t be well-defined and delivered to you. That being said, some product owners and tech leads may have strong ideas for research opportunities. Whatever the case, don’t just passively take on a research project you’re told about, probe deeper through discussions with the team to understand why it matters.</p><blockquote><strong>Focus on understanding product risks, team members’ pains, and the user hypotheses baked into the product. From this, research opportunities will flow.</strong></blockquote><h3>Start scoping and executing research in your second week.</h3><p>At the end of your first week you’ll have a backlog of research opportunities based on your interviews from week one. The backlog will certainly have flaws; it will be incomplete, and based on some misunderstandings. You’ll continuously evolve this backlog as time goes on. However, at this stage you should begin to see some areas where research can likely add value. You may still feel a bit uncertain and uncomfortable in your new workplace, but this is a great time to start doing some research!</p><p>Select a couple of small research projects from your backlog and begin to scope out and execute them. At this stage, you want to choose projects that can be completed in days..</p><p>Be highly collaborative. For each project idea, before starting to collect data, create a one-page summary (something like <a href="https://docs.google.com/document/d/1d0u7HUtk4s8dgNoYecGi4mNx_1uFgXfBxZTzLEZW2rI/edit?usp=sharing">this</a>) and share it with the team to get feedback. Let everyone play a role in shaping the research goals, questions, and methodology. As you run the research, keep track of how comfortable the team is with the research, and be intentional about showing them how it works and demonstrating its value. Help bring team members closer to real users by involving them in running the study and giving them access to the raw data. Set up and scaffold analysis, involving the team members as much as possible. During this time, your team members will learn about research, and you’ll learn about your new workplace and team.</p><p>Try to select an initial set of projects with a good mix of research methods. By the end of the first week, you should start to see a few key methods you’ll probably need regularly. For me, it was usability studies (remote and in-person), surveys, interviews, and focus groups.</p><p>By executing these initial projects early, you’ll learn important lessons on doing research at your new startup. For instance:</p><ul><li><strong>How to access user information</strong>: depending on the organisation, it may be difficult to access user information. At babylon, we take our users’ privacy extremely seriously. This means there is a lot of strict process around accessing users’ information, especially when it relates to medical history.</li><li><strong>Relationship between startup and users</strong>: does your company have a set of die-hard users that will participate in research for free, or will you need to provide incentives? What kinds of incentives are meaningful to the user base?</li><li><strong>Company maturity with regard to user research</strong>: You’ll see in practice how comfortable team members are with user research, which will help you work out how best to collaborate with them.</li><li><strong>Channels for participant recruitment</strong>: You may want to experiment with new, cheaper, or more efficient channels of recruitment to see the trade-offs of each.</li><li><strong>Better, faster methods</strong>: You’ll come into the organisation primed to think of those methods you’re most familiar with. But be open to new ideas from other team members. My UX &amp; Design team have helped stretch my repertoire of methods by pushing me to think beyond my comfort zone.</li></ul><h3>You’ll definitely make mistakes but that’s OK!: Fail and learn fast</h3><p><strong>Some think user research is a slow process. In reality, it can and should be as fast and iterative as any other part of product development, with the same appetite for failing fast. </strong>Your initial research studies will have flaws like you’ll realise some nonoptimal methodological choices or your incentives to recruit participants won’t work well. All important lessons you’ll learn <em>through doing</em>.</p><p>At babylon, we have a strong culture that advocates getting stuck in, failing fast, and evolving. Thanks to my team for being behind me all the way! If babylon sounds like your kind of place, consider joining our team! Check out our <a href="https://www.babylonhealth.com/vacancies/">available roles</a>.</p><p>Give this article a ❤ if it was helpful. Let me know what other lessons you’ve learnt in the comments below or at <a href="/service/mailto:anna.zawilska@babylonhealth.com">anna.zawilska@babylonhealth.com</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b26bce562bb9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My macbook and I wrote a poem together]]></title>
            <link>https://medium.com/@anna.zawilska/my-macbook-and-i-wrote-a-poem-together-3a7b11cd485c?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/3a7b11cd485c</guid>
            <category><![CDATA[poetry]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[life]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[creativity]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Mon, 08 Aug 2016 18:01:55 GMT</pubDate>
            <atom:updated>2016-08-08T18:01:55.280Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GdAlHHPSQ2514WpX8yhKwA.jpeg" /></figure><blockquote><em>the burning fire as the outsiders entered on the narrowing roads<br>barrels stacked in the dark shadows<br>crowds bold and austere</em></blockquote><blockquote><em>their cries unintelligible, bodies creating an island<br>the moon coming up, big yellow bulb</em></blockquote><blockquote><em>the large houses’ windows illuminated<br>children crowded up against the windows<br>babies fallen asleep upstairs, desultory</em></blockquote><blockquote><em>among the thump<br>thump<br>thump</em></blockquote><blockquote><em>a top barrel rolls lurches over<br>clunk<br>on the ground</em></blockquote><blockquote><em>and rolls away to the distance</em></blockquote><p><strong>By Anna Zawilska, and Macbook Serial CO2P…H2, Inspired by Sylvia Plath</strong></p><p>My machine and I wrote the above poem together. The poem is inspired by one of my favourites: Sylvia Plath.</p><p>With just a modest corpus of poetry as a training set, and a couple lines of code, you and your machine can create sweet poetry together.</p><h3><strong>How it works</strong></h3><p>Github repo with code and instructions is here: <a href="https://github.com/AnnaZawilska/inspire-me">https://github.com/AnnaZawilska/inspire-me</a></p><p><strong>The system adapts the idea of autocomplete, to pull out a string of words from a training set of poetry. The words tend to reflect the training data and are rough themes and ideas.</strong></p><p><strong>Then, you use those words as a springboard to write the final poem. You allow the imagery evoked by the machine-generated words to blend with your own thoughts. Images form in the mind, and poetry begins to flow.</strong></p><p>In code, you begin by seeding the system with an initial word (my code uses ‘the’ by default). Then in a loop:</p><p>80% of the time:</p><ul><li>Select a random letter for the next word, and generate a list of possible words to autocomplete</li><li>If some autocomplete suggestions exist, randomly chose one (to cut through words that are very popular)</li></ul><p>20% of the time:</p><ul><li>add a new line. I chose this roughly by taking a sample of a couple of poems and looking at how many of the characters are new line.</li></ul><p>The tool uses this autcomplete system: <a href="https://pypi.python.org/pypi/autocomplete/0.0.104">https://pypi.python.org/pypi/autocomplete/0.0.104</a> It contains a simple “Hidden Markov Model”: Given a current word and the first part of the following word, the tool finds the set of likely second words.</p><p>You train your simple “HMM” with a corpus of works. Generally, the more the better. I scraped around ~10 pieces of Sylvia Plath poetry from internet archives. You want a corpus that has a particular subject matter, so that the concepts pulled out tend to work together to conjure some imagery. The more specific the topics of the poetry in the corpus, the more potentially meaningful the chain of inspirational words your machine generates.</p><p>The inspiration that my machine generated for me for the poem at the top of this article is:</p><blockquote>the<br> outriders pyre<br> narrowed</blockquote><blockquote>time barrels oblique<br> locks adirondacks</blockquote><blockquote>dark</blockquote><blockquote>dew bold generation queen crowd</blockquote><blockquote>vein postman</blockquote><blockquote>big<br> house key end austere fourth fishpond unintelligible island gorse<br> slime other gray big baby yellow island virgins bowel<br> barrels</blockquote><blockquote>fallen<br> chord water connecticut yellow yellow blue upflight<br> window child crowd island</blockquote><blockquote>barrels<br> grownups fallen</blockquote><blockquote>maid nubs<br> water owner end desultory</blockquote><blockquote>island island postman virgins</blockquote><blockquote>last</blockquote><h3><strong>The effect</strong></h3><p>Your poetry is shaped by:</p><ul><li>Your imagery and thoughts</li><li>Your machine’s hardware and software (the random number and letter generation that determine the inspirational words chosen)</li><li>The concepts, themes, and imagery related to the training data</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/450/1*zHYRG3QwLIylp7So-6SIOg.gif" /></figure><h3><strong>Caveat</strong></h3><p>This system is not a good way to train a machine to generate coherent poems by itself. That’s the territory of more complex models with memory like RNN’s. But RNN’s are a lot more computationally complex and usually need more data.</p><h3><strong>Machine and Human Collaborative Poetry Jam</strong></h3><p>This is an example of how even a simple piece of Supervised Learning can synthesise concepts and ideas that humans can use as a springboard for creative tasks.</p><h3><em>If you liked this, click the💚 below so other people will see this here on Medium.</em></h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3a7b11cd485c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Women in tech, heed The Creep]]></title>
            <link>https://medium.com/@anna.zawilska/women-in-tech-heed-the-creep-75542d71247a?source=rss-eea3738ca108------2</link>
            <guid isPermaLink="false">https://medium.com/p/75542d71247a</guid>
            <category><![CDATA[women-in-tech]]></category>
            <category><![CDATA[gender-equality]]></category>
            <category><![CDATA[feminism]]></category>
            <category><![CDATA[diversity]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Anna Zawilska]]></dc:creator>
            <pubDate>Sun, 24 Jul 2016 10:19:05 GMT</pubDate>
            <atom:updated>2016-07-24T12:24:12.958Z</atom:updated>
            <content:encoded><![CDATA[<p>Let’s stop whitewashing gender issues by making them about individual rights.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*Im6YEE77fmPWY0yQ8FWqYA.jpeg" /></figure><p>It’s so easy for the marginalised to slip into inadvertently perpetuating the status quo. In the name of rebranding our cause to give it broader relevance, we dilute or sweep aside the complexities of our lived experience. We map what we really care about into milder, more digestible versions of it.</p><p>And slowly, slowly the core message we need to get across Creeps into a more peripheral position in the conversation.</p><p>The Creep is hard to spot. When you’re entrenched in a particular social framework, the ideas and assumptions it engenders seep into everything you do. It becomes ever harder to disentangle it from reality.</p><p>An important example of The Creep within gender issues in the tech workplace lately, sounds like this: <em>gender inequality is a no brainer, because it’s just about seeing men and women equally; why wouldn’t everyone buy into it?</em><br>In conversations and trainings around unconscious bias, gender issues are framed as an issue that impacts all people; so the crucial action point is to approach each individual afresh, as an individual. End of training. The concept and history of feminism is often reduced and twisted into a movement that calls for all men and women to be treated equally. But Feminism is much more. It’s the complex, tangled, but beautiful reclaim by women of their lives and their bodies. It’s the shining of light onto institutional oppression. To sanitise it, is to undermine those that undertook great personal risk for the collective freedom of women.</p><p>Discussion on this at women in tech events is also an echo chamber for The Creep in wider mainstream media. In a <a href="http://www.unwomen.org/en/news/stories/2014/9/emma-watson-gender-equality-is-your-issue-too">speech</a> as UN Goodwill Ambassador, Emma Watson motivates for a campaign on gender inequality by calling out the ways our modern society also put men in a box, like fathers being less valued as parents. In an often-used <a href="https://youtu.be/-8H7eeMNfLw">TEDx talk</a> on unconscious bias, author and consultant Kristin Maschka calls out the experience of her husband who was at times assumed less able to take care of their child.</p><p>To say that only women are constrained and constricted by society, is also misinformed. As our society begins to accept more complex types of identities, ones that no longer conform to traditional notions of gender or sexuality, for example, it’s true we should try to abandon traditional preconceived notions and support an individual in defining themselves. All the more reason, however, to learn how to have the honest conversations to bring out the broad oppressive forces that act in society, and to tackle them in some prioritised way through policy in the workplace. <strong>Now’s not the time to look <em>through</em> identities and reduce us all to the base unit of an individual.</strong> Technology is eating the world, and in many ways tech employers are leading the trend in creating new kinds of workplaces and models of employment. There’s more weight behind diversity in tech than ever before with <a href="https://techcrunch.com/2016/04/29/chris-sacca-says-theres-a-greed-case-for-diversity/">heavyweights</a> coming forward to make their cases. But heed The Creep.</p><p>The right questions for policy makers in tech companies to ask are something like: “what are the forces in society that work against your identity and true self? How can we be a part of the push back against that?”</p><p>I’m not saying that The Creep is intentional. But in a sense that doesn’t matter. We live in a physical world, we deal with the empirical consequences of even the unintended every day. Leaning on scientific research is crucial here. Gender biases are so sinister in part because they are so subtle yet so mighty. To deny that we have deeply entrenched and significant biases against women in the areas of professional development is to deny a large corpus of scientific work. (For good entry points into this research, I recommend starting <a href="https://hbr.org/2013/09/women-in-the-workplace-a-research-roundup">here</a> and <a href="http://freakonomics.com/podcast/gender-barriers/">here</a>.) The worst thing we can do is to obfuscate them or whitewash biases. We shouldn’t retouch the legacy or the current realities of gender issues to create a prettier picture. We are not your morality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/230/1*zRNV_eVDEzhOZIEw2KvcoA.gif" /></figure><p>We’re better positioned than any generation before us to be transformational. The Creep shows us that we’re not yet free of the status quo and may never be. The Creep exists because we think an honest conversation on gender issues which draws on scientific research, is a risk. But a risk is only one with respect to the norms of the system you’re in.</p><p>The Creep is an unknowing twisting and contrivance of what we truly mean in an attempt to make progress. We dilute our lived experience to make it easier for gatekeepers in the field. But while we’re busy sanitising our cause and skirting around the core issues, we’re missing part of the course forward to becoming gatekeepers.</p><p><em>If you liked this, click the💚 below so other people will see this here on Medium.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=75542d71247a" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>