<?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 Tarun Durga on Medium]]></title>
        <description><![CDATA[Stories by Tarun Durga on Medium]]></description>
        <link>https://medium.com/@tarundurga?source=rss-70d98eca4f6f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*Edo4TrSj5lnFqb45.jpeg</url>
            <title>Stories by Tarun Durga on Medium</title>
            <link>https://medium.com/@tarundurga?source=rss-70d98eca4f6f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:19:40 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@tarundurga/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[The Hidden Cost of Convenience]]></title>
            <link>https://tarundurga.medium.com/the-hidden-cost-of-convenience-72ce27ce0b99?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/72ce27ce0b99</guid>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[lifestyle]]></category>
            <category><![CDATA[consumerism]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Mon, 29 Sep 2025 12:34:17 GMT</pubDate>
            <atom:updated>2025-09-30T05:30:12.825Z</atom:updated>
            <content:encoded><![CDATA[<p>When my sister and I were kids, we’d go shopping with our dad, and it was always traumatic for him. We both wanted something — it didn’t matter what, it could be a pencil — but we wanted it. While I sulked, my sister, the more precocious one at that age, would throw a full-blown 8.5 Richter tantrum. Dad, being a gentle soul, would give in. Over dinner, he’d tell us: <em>“If you want something, wait. Sleep on it for a few days. If you still want it after that, then get it.”</em></p><p>I feel the weight of those words today. And I find it even harder now to follow his advice. I think we all do. There is simply too much convenience. You can get what you want by barely thinking about it — it’s all just a swipe or a click away. Got a craving for chocolate brownie ice cream at 10 p.m.? A couple of taps, and it’s at your door by 10:15. Can’t think of an apology? GPT will frame one for you. What can wait, doesn’t. What needs your presence, can’t have it. Lazy use of AI feeds words into our heads and leads to an impersonal sameness. Epiphany is lost. Serendipity is lost. Agency is lost. Such beautiful words, drained of their meaning. The hidden cost of convenience is a growing sense of meaninglessness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gq4gbMwjT5N7gaJVM5vrHw.png" /><figcaption>Cult of Now</figcaption></figure><p>My wife and I have made a conscious decision not to give in to quick commerce. It comes at the cost of exhausted riders, shaky unit economics, and reputational blowback. E, S, and G are all compromised for something that can easily wait a few days.</p><p>But there’s more. When you buy something today, the whole world pays for it. Resources are finite, as are human effort and time. Our appetite isn’t. The hidden cost of convenience is borne by workers, by the planet, by businesses — and eventually by ourselves. Every time you buy a single-use bottle of water, you add to the debt that will be carried by future generations. Just carry your own bottle.</p><p>Designing for convenience has another unintended side effect: fragility. Supply chains built for hyper-efficiency collapsed during COVID when a single missing semiconductor halted entire industries. Buy Now, Pay Later (BNPL) has wrecked the finances of people who no longer had to wait before buying the latest iPhone. Auto-saving everything to cloud servers has created digital dump yards that echo the ones in the real world — raising costs, reducing efficiency, and leaving behind a heavy environmental footprint.</p><p>I’m not advocating a survivalist way of life. Convenience is good. But so is friction. It gives us pause. Instead of designing friction out, we need to mindfully build it into our products and services. It’s a balancing act that requires us to carefully distinguish between wants and needs.</p><p>Friction builds resilience and appreciation. Meditation can be a drag — until it isn’t. A meditation app on my phone rewards users with convenient access to different forms of breathwork and mindfulness practices. But it also rewards persistence, encouraging people to return day after day, even for just 10 minutes.</p><p>We value what takes effort to build. LinkedIn profiles can be a pain to create, but once complete, the network effect compounds over time. Supply chains designed with slack are more robust in times of uncertainty.</p><p>Friction also fosters awareness and trust. Reading the environmental footprint on product packaging helps us make better decisions. Two-factor authentication may irritate, but it reassures. So does consent before auto-renewal. On the other hand, the tiny text in legalese on insurance forms is not good friction — it’s terrible. There’s no excuse for it.</p><p>Deciding between friction and convenience is not rocket science. It’s about intentionality. It’s about asking ourselves when to apply which. Here are five How-Might-We questions to consider the next time you are working on UX, a product, a service — or absolutely any kind of design:</p><ol><li>How might we design convenience that reinforces deliberate choices rather than impulsive ones?</li><li>How might we introduce friction that deepens learning, growth, or sustainable behaviour instead of getting in the way?</li><li>How might we make the invisible costs of convenience — social, environmental, financial — visible at the right moment?</li><li>How might we create experiences that feel rewarding in the moment while compounding value for users and businesses over the long term?</li><li>How might we frame certain inconveniences as intentional features that add meaning, appreciation, or responsibility?</li></ol><p>We need to stop thinking of convenience as a universal attribute and peel it off layer by layer. We might realise that what feels like convenience to one person may feel like exclusion to another. At the end of the day, everything is interrelated, and it’s the quality of our relationships that shapes the texture of our lives.</p><blockquote><em>“When you drink a cup of tea, you can see the whole universe in it — the rain, the soil, the sunshine, the farmer, the picker. In that moment you are in touch with all that is.”</em> — Thích Nhất Hạnh.</blockquote><p>The same applies to design. Every choice, every layer of friction or convenience, carries the whole system within it. In a world drowning in dopamine hits and degrading ecosystems, we need to be able to take step back and ask ourselves — what are we enabling by removing friction?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=72ce27ce0b99" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Trying to pull these networks into centralised platforms won’t work-  meet them where they are.]]></title>
            <link>https://tarundurga.medium.com/trying-to-pull-these-networks-into-centralised-platforms-wont-work-meet-them-where-they-are-966d0313711c?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/966d0313711c</guid>
            <category><![CDATA[ecology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[biomimetics]]></category>
            <category><![CDATA[systems-thinking]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Mon, 18 Aug 2025 07:27:06 GMT</pubDate>
            <atom:updated>2025-08-18T07:27:06.005Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7_Pik4UPzWjPJvzAk4cnhA.png" /></figure><p>Much of what holds our world together doesn’t sit in a boardroom or a control centre. It lives in the countless small connections between people, places, and communities. Think of neighbourhood WhatsApp groups coordinating medical help during COVID, Kudumbashree women’s collectives in Kerala running community kitchens, or volunteers mapping flood relief needs in Assam. None of these depend on a single authority. They’re examples of <strong>distributed networks</strong> — systems where coordination and resilience emerge from many small links.</p><p>You can usually spot them if you know what to look for. They work without a central hub, and yet useful things keep happening. They flow many-to-many instead of top-down. They adapt when one node fails: if one relief truck can’t reach a village, another volunteer steps in with a different solution — organising a local kitchen, setting up a temporary shelter, or coordinating medical help — whatever the situation most needs at that moment. And the part you see — the event, the funder’s logo, the press release — is usually just the surface. The real work sits underneath, in trust, relationships, and the quiet exchange of resources.</p><p>Nature offers the clearest metaphor. A mushroom is just the fruiting body. The real action is in the vast mycelial web underground, routing nutrients, relaying signals, adapting to changing conditions. Peter Wohlleben, author of the <em>The</em> <em>Hidden Life of Trees</em>, describes how these webs allow trees to share sugars, send warnings about pests, and nurture saplings in the shade, turning a forest into a social community where survival is collective rather than individual. He likens them to an underground internet, a living glue that binds the forest together and keeps information and resources flowing across the whole ecosystem.</p><p>In our context, distributed human networks function the same way — sharing what is needed, when it is needed, quietly and without PR explosions — so that the community becomes stronger together than any single part could on its own. The vaccination drive that “suddenly” reaches a million people? It’s often the result of weeks of invisible coordination between community health workers, self-help groups (SHGs), and local supply chains quietly doing their part.</p><h3>What makes distributed networks special</h3><p>A few qualities stand out once you notice them.</p><p>They are resilient: if one strand breaks, another carries the load, much like ride-sharing drivers stepping in when public transport halts.</p><p>They are adaptive: decisions happen close to context, such as fisherfolk adjusting catch practices faster than central regulators when tides shift.</p><p>They coordinate through emergence: think of open-source software, where no single leader dictates every line of code, yet coherent products emerge.</p><p>They favour ongoing, mutual relationships — like SEWA’s model of women supporting each other with credit, childcare, and healthcare over decades — rather than one-off transactions.</p><p>And by their very nature, they remain less visible than centralised systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AZGHrx-7Quj-bGsaD2nxjQ.png" /><figcaption>SHGs as the invisible layer.</figcaption></figure><h3>What they need most</h3><p>Here’s the tension: the very features that make distributed networks strong also make them harder to support. They need visibility in the sense of being discoverable. The many local Self Help Groups (SHGs) that fed migrant workers during India’s lockdown, for instance, rarely made it into policy conversations despite their scale.</p><p>They need connectivity and cooperation that goes beyond funding. During Kerala’s floods, informal WhatsApp groups of boat operators and local officials saved lives simply by sharing who had fuel and who had access.</p><p>They need local grounding: a beautifully designed portal in English is useless if the farmers it hopes to serve rely on radio in their dialect. They need a shared language, because policymakers may talk about “resilience” while villagers talk about “coping,” and the two don’t always meet.</p><p>They need recognition as equals: India’s SHG-bank linkage programme works precisely because it doesn’t treat women’s groups as beneficiaries but as partners in formal credit.</p><p>And they need lightweight infrastructure — simple, mobile-friendly ways to coordinate, like IVR hotlines or Google Sheets — rather than complex systems that collapse under their own weight.</p><h3>Meeting them where they are</h3><p>The mistake is to pull these networks into our platforms and processes. It rarely works. A healthier approach is to meet them where they are. That means funding coordination, not just outputs — think of paying for translators, conveners, or local data managers who keep the threads alive. It means building on tools already in use: M-Pesa in Kenya or UPI in India scaled because they fit with existing habits, not because they introduced brand-new ones. It means accepting that impact rarely comes through a single funnel: microcredit, peer training, and local barter often work together.</p><p>The point is to strengthen the connective tissue — and curb the urge to centralise.</p><h3>Where AI fits in</h3><p>AI has a role here, but not as the star. Used carefully, it can act like mycelium: mostly unseen, strengthening flows at the edges.</p><p>It can help surface hidden work by scanning community radio transcripts, local-language news, or district-level NGO reports that don’t usually reach policymakers.</p><p>It can strengthen connectivity by translating between a municipal circular and a grassroots group’s WhatsApp update, ensuring both sides understand each other.</p><p>It can support local grounding by offering voice-based tools in regional languages, like WhatsApp bots that respond in Kannada or Assamese. It can create a shared language by producing plain-speak briefs — turning a dense climate policy into a two-minute audio clip that a farmer or councillor can digest.</p><p>It can offer lightweight infrastructure: auto-generated meeting notes, chatbots that answer basic queries, or dashboards that both a village NGO and a corporate CSR team can co-use.</p><p><em>Across sustainability, communications, policy, and green commerce, the principle is the same. AI should strengthen what is already happening.</em></p><p>In sustainability, it can map who is piloting which solution so duplication is avoided. In communications, it can build glossaries so funders and activists stop talking past one another. In policy, it can help sift citizen feedback to highlight what’s urgent. In green commerce, it can shift the focus away from nudging more consumption with one-off discounts, and towards amplifying models that encourage repairing, sharing, reusing — rewarding customers and companies that choose to buy less but better, and to extend the life of what already exists.</p><h3>A final nudge from fungi</h3><blockquote>Mycelial networks remind us of a few things: build strong but hidden connective tissue; let flows self-organise; favour partnerships over transactions; design redundancy; treat proof as seasonal fruit, not the whole tree; and scale through spores — small, lightweight patterns that others can pick up and grow.</blockquote><p>Most progress doesn’t come from the spotlight. It comes from networks that don’t seek attention but hold things together day after day. Our role — whether in NGOs, companies, or government — is to notice them, strengthen their threads, and respect their ways of working.</p><p>AI won’t fix everything. But with humility and care, it can help these networks do what they already do best: quietly hold the forest together so that, in time, more fruit can emerge.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=966d0313711c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Start Small, Win Fast! Building AI Confidence in Marketing Teams]]></title>
            <link>https://tarundurga.medium.com/start-small-win-fast-building-ai-confidence-in-marketing-teams-4435ac9bfc42?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4435ac9bfc42</guid>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[digital-transformation]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Thu, 07 Aug 2025 16:00:22 GMT</pubDate>
            <atom:updated>2025-08-07T17:26:31.822Z</atom:updated>
            <content:encoded><![CDATA[<h3>Start Small, Win Fast: Building AI Confidence in Marketing Teams</h3><p>AI is everywhere—on every blog, in every boardroom, and in almost every vendor pitch. If you&#39;re a marketing head right now, chances are someone on your team has already said, “Should we be using AI for this?”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DpMAR1aPLGFrXLZfsTfBOA.png" /><figcaption>Start your AI journey with small experiments.</figcaption></figure><p>The answer is yes. But not in the way you might think. You don’t need to rebuild your marketing strategy from scratch. You don’t need to hire a data science team. And you certainly don’t need to roll out a 12-month AI transformation plan.</p><p>What you need to do is start with +AI—small, smart add-ons that plug into your existing marketing engine and make it run faster, smoother, and smarter.</p><h3>Why Marketing Teams Can’t Ignore AI Anymore</h3><p>Marketing today is too fast and too fragmented to run only on human horsepower. AI gives you leverage where it hurts the most:</p><ul><li>Too much content to create – blogs, emails, ad copies, captions, all yesterday.</li><li>Too many touchpoints to manage – WhatsApp, email, Insta, SMS, web, all needing consistency.</li><li>Too many decisions to make – what campaign to run, what to say, whom to say it to, when.</li></ul><p>Add AI, and you move from knee jerk reactions to smart, proactive action. Let’s see how.</p><h3>The +AI Progression Ladder: One Win at a Time</h3><p>Think of +AI as adding a motor to a bicycle—it helps you move faster, but you still steer. The beauty is, you don’t need to change your path. Just start climbing.</p><p>Here’s what a +AI journey can look like in your marketing function:</p><h4>Step 1: Personal Productivity +AI</h4><p><em>For individual team members</em></p><ul><li>Use ChatGPT, Gemini, or Claude to write blog drafts, social posts, or email copies.</li><li>Summarise long documents, competitor news, or performance reports instantly.</li><li>Turn bullet points into polished content for campaigns or presentations.</li></ul><p><strong>Goal:</strong> Save 30–40% of time on routine creative and analysis tasks.</p><h4>Step 2: Team Workflow +AI</h4><p><em>For internal process boosts</em></p><ul><li>Use Google Sheets with AI plugins (like ChatGPT in Excel) to clean campaign data.</li><li>Automate monthly performance summaries using Notion AI or Airtable AI.</li><li>Build simple prompt templates for brand voice or campaign tonality—usable by interns or execs.</li></ul><p><strong>Goal:</strong> Improve consistency and speed across the team, reduce repetitive work.</p><h4><strong>Step 3: Channel-Specific +AI</strong></h4><p><em>For targeted wins</em></p><ul><li>Use Mailmodo with AI subject line and copy generation for email campaigns.</li><li>Use Writesonic or Narrato for SEO blog outlines and long-tail keyword mapping.</li><li>Use Adobe Firefly or Canva AI for fast visual content versions in regional languages.</li></ul><p><strong>Goal:</strong> Increase channel ROI with AI-powered enhancements.</p><h4><strong>Step 4: CRM &amp; Campaign +AI</strong></h4><p><em>For smarter targeting and nurturing</em></p><ul><li>Use Salesforce’s Einstein GPT to write follow-up messages based on lead profiles.</li><li>Auto-summarise prospect conversations and engagement history for sales handovers.</li><li>Score leads based on behaviour and trigger micro-journeys automatically.</li></ul><p><strong>Goal:</strong> Boost lead conversion and shorten nurture cycles.</p><h4><strong>Step 5: Decision Support +AI</strong></h4><p><em>For insights and planning</em></p><ul><li>Ask AI tools to evaluate campaign performance and suggest what to double down on.</li><li>Build dashboards that answer questions like “Which channel brought in most high-LTV users?”</li><li>Test messaging using AI-powered customer simulations.</li></ul><p><strong>Goal:</strong> Move from guesswork to guided planning.</p><h3><strong>Okay, But What’s AI+ Then?</strong></h3><p>By the time you’ve hit Step 5, you’ll start asking bigger questions:</p><p>Why are we still briefing campaigns manually? Can’t AI build the first draft of our monthly calendar? What if we used customer data to personalise the entire journey, not just the email subject?</p><p>That’s when you start thinking about AI+—redesigning how marketing works, with AI at the centre.</p><p><strong>What Needs to Be in Place Before You Begin</strong></p><p>Before you go all in, make sure you’ve got your basics sorted:</p><ul><li><strong>Clean CRM data</strong> – junk in, junk out.</li><li><strong>Defined brand voice</strong> – so your AI outputs sound like you.</li><li><strong>Internal champions</strong> – people who can test, learn, and teach others.</li><li><strong>A willingness to try</strong> – even if it feels imperfect at first.</li></ul><p>For example, if you’re using Salesforce, start by integrating the free tier of Einstein. Even just auto-drafting meeting notes or follow-up mails can unlock hours per week and better closure rates.</p><p><em>But you don’t need to start there.</em> In fact, the best way to get to AI+ is by mastering +AI — one confident step at a time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2nK6NiKIIFpiS-Ind-IKWw.png" /></figure><h3><strong>The Future Isn’t Optional Anymore</strong></h3><p>Teams that learn to ride the AI wave will ship faster, test smarter, and resonate deeper. Teams that don’t will fall into the noise.</p><p>This is much more than chasing the next tech trend. It’s about giving your team the clarity, confidence, and control to stay ahead in a chaotic world.</p><p>If this sounds like a conversation worth having, I’d love to help. Let’s start small and build momentum. Let’s get +AI, right.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4435ac9bfc42" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[GenAI Is an Energy Hog. I Still Think You Should Use It.]]></title>
            <link>https://tarundurga.medium.com/genai-is-an-energy-hog-i-still-think-you-should-use-it-a4c97482356f?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/a4c97482356f</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[esg]]></category>
            <category><![CDATA[responsible-ai]]></category>
            <category><![CDATA[sustainability]]></category>
            <category><![CDATA[gen-ai-for-business]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Thu, 24 Jul 2025 11:44:51 GMT</pubDate>
            <atom:updated>2025-07-24T12:16:21.211Z</atom:updated>
            <content:encoded><![CDATA[<h3>GenAI is an Energy Hog. I Still Think You Should Use It.</h3><p>The sustainability head of a financial services firm told me that GenAI for sustainability was an oxymoron. What you’re about to read draws from that chat. Straight up, he is right, but that is just one side of it.</p><p>As GenAI tools like ChatGPT make their way into boardroom strategies, financial reports, and sustainability workflows, this question hangs in the air: Can a technology with such a large environmental footprint really be part of the sustainability conversation?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W8IkkhtsQ4f6JYsTONuRmg.png" /><figcaption>Energy Guzzler or Humanity’s Saviour?</figcaption></figure><p>This piece is structured as a dialogue between a <strong>Sceptic</strong> and an <strong>AI Evangelist</strong>. The Sceptic voices common concerns from emissions to water use to greenwashing — while the AI Evangelist offers a grounded counterpoint. Rather than debate for the sake of it, the two explore nuance: when GenAI is worth the cost, when it’s not, and how sustainability-minded professionals can use it without compromising their values.</p><p>If you’ve found yourself torn between urgency and ethics, innovation and impact — this conversation is for you.</p><p><strong>Sceptic:</strong><br>There’s been a lot of talk about the environmental cost of generative AI. I keep seeing claims that using something like ChatGPT is worse than flying — is it true?</p><p><strong>Evangelist:</strong><br>It’s a claim that grabs attention, but it needs context. Training a large model like GPT-4 does consume a considerable amount of energy, sometimes equivalent to what a small town might use in a few weeks. But it refers to the <strong>training phase</strong>, which happens once.</p><p>What most people are concerned about, and rightly so is what happens every time we use these tools. That’s called inference. And the energy use per prompt there is far lower — closer to running a search engine query or streaming a short video clip.</p><p><strong>Sceptic:</strong><br>So the bulk of the impact is upfront, not with every question I ask?</p><p><strong>AI Evangelist:</strong><br>Exactly. And we also need to factor in the direction the technology is moving in. Newer models are being built to be smaller and more efficient — think DeepSeek, Claude, Phi-3, Mistral. There’s active competition around reducing size and energy use while keeping performance high.</p><p>And here’s where second-order thinking helps. Instead of asking only “How much energy does AI use?” we should also ask: “What does it replace?”</p><p>If AI can cut hours of manual work, reduce meetings, streamline complex modelling — there’s a chance it displaces emissions elsewhere. And that trade-off needs to be part of the equation.</p><p><strong>Sceptic:</strong><br>That’s a helpful shift in perspective. Still, the energy use is real — and growing. Should we be worried about where this is heading?</p><p><strong>AI Evangelist:</strong><br>It’s right to be concerned. But also worth remembering that most transformative technologies — from cloud computing to mobile phones — started off inefficient. What matters now is how we guide this one: how we build, where we deploy, and what problems we choose to solve with it.</p><p><strong>Sceptic:</strong><br>Let’s talk about what’s behind the headlines. Why does GenAI use so much energy and water in the first place?</p><p><strong>AI Evangelist:</strong><br>There are two main stages: <strong>training</strong> and <strong>inference</strong>. Training is the heavyweight — that’s when vast amounts of data are processed through neural networks over weeks or even months. It’s computationally intense and often powered by energy-hungry GPUs in large data centres.</p><p>Inference is what happens when you use the model — typing in a prompt and getting a response. That’s lighter, but when done millions of times a day across the world, it adds up.</p><p>Data centers use water for cooling. If the energy source is fossil-fuel-heavy, that heat becomes an issue — so cooling systems circulate large volumes of water. But when you read that “a single AI query consumes 500 ml of water,” it’s important to ask, <em>“Compared to what?”</em></p><p><strong>Sceptic:</strong><br>Yes, I saw that stat recently — and I didn’t quite know what to do with it.</p><p><strong>AI Evangelist:</strong><br>That’s the challenge with how some of these numbers are presented. They sound dramatic on their own — but without context, they’re misleading. This is a classic case of <strong>“base rate neglect.”</strong> You’re given an eye-catching stat, but not the typical rate or the range.</p><p>For example, boiling a kettle uses water and energy too. So does watching Netflix, or training a search algorithm. We’re not saying GenAI is impact-free, far from it, but to think critically, we need to anchor the stat. What kind of data centre? Powered by what energy mix? Used for what purpose?</p><p><strong>Sceptic:</strong><br>Isn’t this just another hype cycle? Big Tech racing to slap a sustainability sticker on everything?</p><p><strong>AI Evangelist:</strong><br>We’ve seen this pattern before — with blockchain, the cloud, even “carbon neutral” streaming platforms. New technology shows up, gets industrial-sized fast, and then the sustainability pitch follows. Usually louder than the scrutiny.</p><p>So yes, there’s green spin in the AI space. It shows up in vague commitments, selective reporting, and high noise-low substance “AI for good” campaigns. But that doesn’t mean all uses are hollow — or that responsible deployment isn’t possible.</p><p><strong>Sceptic:</strong><br>But if everyone’s using it — for everything — aren’t we back in that familiar place? Shared resources, individual gains, collective costs?</p><p><strong>AI Evangelist:</strong><br>Exactly. That’s the<strong><em> Tragedy of the Commons</em></strong> in action. Everyone wants to use the best tools — to optimise, communicate, innovate — and if AI promises all that, adoption spreads fast. But if we don’t apply <strong>intelligent restraint</strong>, we risk burning out shared infrastructure: energy, water, credibility.</p><p>That’s why I believe we should stay sceptical — and still engage. Not with blind enthusiasm, but with intention.</p><p><strong>Sceptic:</strong><br>And what does “intelligent restraint” look like in practice?</p><p><strong>AI Evangelist:</strong><br>It means asking: Does this need to be generated? Can a smaller model do the job? What’s the marginal value of this output — and who benefits?</p><p>It’s a mindset shift. Just because we can doesn’t mean we should. But let’s also not throw the whole toolbox away because some people are using it to build sandcastles.</p><p><strong>Sceptic:</strong><br>Let’s talk possibilities for a while. Can GenAI actually help sustainability efforts in a meaningful way?</p><p><strong>AI Evangelist:</strong><br>Generative AI isn’t a climate solution on its own. But it can be a force multiplier in areas that already matter. Here’s where it’s already making a difference:</p><ul><li><strong>Sustainability reporting</strong>: GenAI is speeding up ESG disclosures, identifying gaps, and even drafting narrative sections aligned to frameworks like GRI and BRSR. It doesn’t replace the human, but it shortens the grind — especially for smaller teams.</li><li><strong>Risk modelling</strong>: It’s helping climate scientists and insurers simulate complex scenarios — like flood risks, heatwaves, or supply chain breakdowns. Some models are even being used to translate technical forecasts into plain language for local governments.</li><li><strong>Energy efficiency</strong>: In buildings, factories, and data centres themselves, AI is optimising cooling, power use, and predictive maintenance — cutting energy use at scale.</li><li><strong>Citizen engagement</strong>: NGOs are using GenAI to translate complex environmental issues into locally relevant content — in regional languages, with culturally resonant metaphors.</li></ul><p><strong>Sceptic:</strong><br>Makes me wonder if it would be more responsible to just avoid using GenAI unless it’s absolutely necessary.</p><p><strong>AI Evangelist:</strong><br>Yes, but we need to be careful with the framing. Because “just avoid it” sounds simple — and sustainability rarely is. If we zoom out a little, what we should be really asking is: <em>What’s the opportunity cost of not using AI?</em></p><p>If a team spends weeks manually consolidating ESG data across spreadsheets, when a GenAI tool could handle the bulk in minutes — what happens to the time they could’ve spent analysing trends, improving impact, or engaging stakeholders?</p><p>Sometimes using a high-emission solution now unlocks longer-term benefits we’d miss if we waited for the perfect solution.</p><p><strong>Sceptic:</strong><br>Let’s say we take all this seriously. What does a more sustainable GenAI ecosystem actually look like in practice?</p><p><strong>AI Evangelist:</strong><br>It starts upstream — with how models are trained, where they’re hosted, and what we prioritize. Here are a few concrete shifts that would make a real difference:</p><ol><li><strong>Cleaner training environments</strong><br> Using data centres powered by renewable energy should be the norm, not the exception. Some cloud providers already claim high percentages — but the true carbon footprint depends on where and when energy is drawn. Transparency matters here.</li><li><strong>Smaller, smarter models</strong><br> Not every task needs a GPT-4-sized hammer. Lightweight, open-source models — fine-tuned for specific jobs — can reduce both energy use and processing time. Think precision over power.</li><li><strong>Transparency in reporting</strong><br> If companies start disclosing the carbon and water cost of model training, we can compare systems, hold claims accountable, and make informed choices. This is basic sustainability hygiene — and we’re not there yet.</li><li><strong>Shared infrastructure for shared benefit</strong><br> Instead of every company training its own massive model, collaborative frameworks — like open science labs, public-private partnerships, or pooled datasets — can reduce redundancy and distribute both impact and access.</li><li><strong>Use-case restraint and design ethics</strong><br> Lastly, it comes down to what we <em>choose</em> to build. Models built to support regenerative agriculture or disaster preparedness carry a very different weight than those generating synthetic spam. The why matters as much as the how.</li></ol><p><strong>Sceptic:</strong><br>So if I’m someone working in sustainability — or just trying to act responsibly — what questions should I ask before bringing GenAI into my work?</p><p><strong>AI Evangelist:</strong><br>Start with intent. Not whether AI is “good” or “bad,” but whether it’s fit for purpose — and whether it helps you do something better and more meaningfully, not just faster.</p><p>Here are five questions I keep coming back to:</p><ol><li>What problem am I solving — and is GenAI the right tool for it.</li><li>What is the environmental cost — and is it proportionate to the value created?</li><li>Can I use a smaller, more efficient model — or run it locally?</li><li>What assumptions is the model making — and who might be excluded?</li><li>Will this use of GenAI help build capacity — or just automate for the sake of it?</li></ol><p>It’s not about perfection. It’s about using emerging tools with eyes open — and with values intact. Sustainability is as much about avoiding harm as it is about designing for a future we want to live in.</p><p><em>Sources &amp; References:</em></p><ul><li>MIT News (2025). Explained: Generative AI’s Environmental Impact. <a href="https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117">https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117</a></li><li>Nature India (2024). The water footprint of AI. <a href="https://www.nature.com/articles/d44151-024-00024-8">https://www.nature.com/articles/d44151-024-00024-8</a></li><li>ScienceDirect (2024). Energy-efficient design of transformer-based models. <a href="https://www.sciencedirect.com/science/article/pii/S0925231224008671">https://www.sciencedirect.com/science/article/pii/S0925231224008671</a></li><li>Sopra Steria (2024). How to use AI for an ethical and sustainable future. <a href="https://www.soprasteria.co.uk/insights/blogs/details/how-to-use-ai-for-an-ethical-and-sustainable-future">https://www.soprasteria.co.uk/insights/blogs/details/how-to-use-ai-for-an-ethical-and-sustainable-future</a></li><li>Gramener (2023). Generative AI for ESG: A Sustainability Roadmap. <a href="https://blog.gramener.com/generative-ai-for-esg-sustainability-roadmap/amp/">https://blog.gramener.com/generative-ai-for-esg-sustainability-roadmap/amp/</a></li><li>Microsoft &amp; UC Riverside Study (2023). Making AI environmentally sustainable. <a href="https://arxiv.org/abs/2304.03271">https://arxiv.org/abs/2304.03271</a></li><li>Stanford HAI (2024). Foundation Model Transparency Index. <a href="https://hai.stanford.edu/news/foundation-model-transparency-index">https://hai.stanford.edu/news/foundation-model-transparency-index</a></li><li>World Economic Forum (2024). Responsible AI Playbook. <a href="https://www.weforum.org/reports/responsible-ai-playbook/">https://www.weforum.org/reports/responsible-ai-playbook/</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a4c97482356f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Could Take Your Job — Or Not.]]></title>
            <link>https://tarundurga.medium.com/ai-could-take-your-job-or-not-9a6d3e7ffb32?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/9a6d3e7ffb32</guid>
            <category><![CDATA[business-coaching]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[executive-coaching]]></category>
            <category><![CDATA[digital-transformation]]></category>
            <category><![CDATA[future-of-work]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Wed, 23 Jul 2025 13:53:35 GMT</pubDate>
            <atom:updated>2025-07-23T13:53:35.087Z</atom:updated>
            <content:encoded><![CDATA[<h3>AI Could Take Your Job — Or Not.</h3><p>AI is already transforming the very nature of work. It’s no longer a question of “if” but “how.” Can AI take your job? It’s possible, and it depends on your response to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/962/1*J4GeNYFztGlfzGXg8wQtQQ.png" /></figure><p>Most people think adopting AI means using tools to do the same work faster or cheaper. That’s a short-term gain, not a long-term advantage. If all you’re doing is automating your own work, you’re probably making yourself redundant without realising it. The people who will thrive in the AI era are the ones who learn to think with it — not just use it. Let me qualify that further — think with it, while rethinking we work.</p><p><strong>That shift makes all the difference.</strong></p><p>Thinking with AI is about more than prompts or outputs. It’s about asking better questions, interpreting ambiguous signals, and guiding outcomes with intentionality. It’s about knowing when to rely on AI, when to override it, and how to design solutions that blend machine capabilities with human depth. This is the new skillset. And it’s not optional.</p><p>While tools, interfaces, and hacks will keep evolving — judgment, empathy, strategic clarity, and contextual thinking will endure. These are the uniquely human skills that machines can’t replicate. In fact, they matter more than ever in a world increasingly shaped by automation. AI is not something to fear — or ignore.</p><p><strong>AI excels at managing complexity. You excel at understanding it.</strong></p><p>You bring memory, meaning, emotion, and experience to the table — all the things that make decision-making human. You know what it feels like to doubt, to hope, to imagine outcomes and weigh trade-offs. That’s what gives you an edge. When you integrate that awareness into your work with AI, you unlock a different kind of potential.</p><p><strong>This is the mindset I help leaders build.</strong></p><p>I work with senior executives and their teams to go beyond surface-level AI adoption. We co-build real-world thinking systems. We identify use cases, frame sharper decisions, and imagine smarter futures — using AI not as a crutch, but as a catalyst for redesigning workflows and architecting new outcomes. The goal isn’t just to use tools. It’s to design new ways of thinking and leading, augmented by intelligent systems.</p><p><strong>Sail where others drift.</strong></p><p>You are the captain of your ship. You have the experience and instinct. Think of me as your navigator. Helping you chart smarter courses. Let’s make AI your unfair advantage. If you’re curious about what’s possible, let’s talk. I’ll be happy to send across use cases in your line of work for you to explore.</p><p>Visit <a href="https://tarunonai.com/">TarunOnAI.com</a> to book a free discovery session.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a6d3e7ffb32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Book Whisperer — AI Assistant for Bibliophiles]]></title>
            <link>https://tarundurga.medium.com/the-book-whisperer-ai-assistant-for-bibliophiles-766e62d5a97e?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/766e62d5a97e</guid>
            <category><![CDATA[ai-for-good]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[book-recommendations]]></category>
            <category><![CDATA[books]]></category>
            <category><![CDATA[writing-prompts]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Tue, 22 Jul 2025 07:23:20 GMT</pubDate>
            <atom:updated>2025-07-22T07:23:20.064Z</atom:updated>
            <content:encoded><![CDATA[<h3>The Book Whisperer — AI Assistant for Bibliophiles</h3><p>Here’s a very cool GenAI use case for lovers of reading- an AI book discovery assistant that knows what you’ve read and loved, and recommends books that sing to your soul. These suggestions are informed by your reading history and preferences, and then gradually grow more eclectic — still tied to the thread of your experience and interests. It’s unlikely you’ll come across something already on your shelf, and more than likely that you’ll enjoy it immensely.</p><p>It draws from a deep sense of nostalgia. I remember, in the early years of the internet, discovering my favourite books through pure serendipity. There were no recommendation engines at the Sunderbai Hall Book Fair. We had to slowly sift through long rows of new and weather-worn books. The joy was in trying to get a sense of the story while relying on memory and excitement. I can still smell the air! It took half a day of beautiful browsing, and I carried back a ton of books, and anticipation.</p><p>My favourite people were the all-knowing booksellers at Lotus and Strand Bookshops in Mumbai, and Fact &amp; Fiction in Delhi. These guys were bona fide bibliophiles — they knew their stuff, and over time, they got to know what you liked too. Their recommendations were almost always spot on. Virat Chandhok of Lotus is thankfully still around, working his magic at White Crow in Mumbai. It’s always a pleasure to chat about what he’s reading, and flip through what he believes I should read. He knows my reading history better than Amazon ever could.</p><p>Contrast this with how we discover books today. Amazon and other marketplaces suggest titles you’ve browsed, not bought. You’re shown what others have picked up, plus a barrage of sponsored sales. It works, sure — but it feels distant. Pushy. Most of the old bookshops have shut down. And in my experience, many booksellers are just sellers — put them in a sanitaryware store, and they’d fit right in. That bookishness is missing. Bookstagrammers exist, of course — but you have to swim through an ocean of droll before you find someone who shares your interests.</p><p>I know I might be overthinking this, and also wildly generalising — but like all bibliophiles, I have strong feelings about what I read and how I find it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZGp2ZTlWKeJTUMKEo1E_6A.png" /><figcaption>Created by Dall-E, but looks like a cousin of Ajit Vikram Singh of Fact &amp; Fiction in Delhi</figcaption></figure><p>Anyway — so here’s what I’ve done. I downloaded a CSV of all my Kindle purchases — going back to around 2013. I’ve done the same with my Goodreads list (fortunately, I’ve been disciplined about updating it). My wife (also a bibliophile) and I exchange notes and compete in the Goodreads Challenge each year. She wins. I get by. As a result of both these CSVs, I now have a consolidated list of 2000+ books spanning 12–13 years. That’s a lot of history. Just what an LLM loves.</p><p>I’ve created a custom GPT called The Book Whisperer — because what is life without some drama? The knowledge base includes those CSVs. The prompt instructs it to adopt a personality reminiscent of my favourite booksellers — I can almost hear moody blues playing at Fact &amp; Fiction. I ask it questions like: “Suggest something that feels like the colour crimson,” or “I loved this book because of XYZ — what else should I read?” It then recommends three titles that aren’t on either of my lists, and are closely aligned with what I love. It knows this from the data in the CSVs — it’s done a sentiment analysis. And if I’m feeling dangerous, I can override that too.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3-wIekiVWp3dNMt5rXn8TQ.png" /><figcaption>3 levels of recommendations — Safe Bet, Stretch &amp; Wildcard</figcaption></figure><p>Bottom line: a personalised set of recommendations I can trust, in a voice that feels familiar, and articulated in a way that truly speaks to me.</p><p>I’ve been using The Book Whisperer for a few days now, and it has been delightful. I’d love to share this old-skool nostalgia with you. <a href="http://tarundurga@gmail.com/">Email</a> me for an editable copy of the prompt so you can run it on your own GPT. You don’t need years of history — just list your favourite books in a spreadsheet with the title, author, and genre. Your AI will learn more as you engage with it.</p><p>You can experience The Book Whisperer too — but you’ll need a ChatGPT Plus subscription that enables access to custom GPTs. You’ll get to see some seriously quirky recommendations — and a peek inside my head. Shoot me a message, and I’ll send you the link.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=766e62d5a97e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Wondering where to start with AI? Read this first.]]></title>
            <link>https://tarundurga.medium.com/wondering-where-to-start-with-ai-read-this-first-db1c096cee18?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/db1c096cee18</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[ai-framework]]></category>
            <category><![CDATA[human-centered-ai]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[ai-innovation]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Sun, 20 Jul 2025 13:50:29 GMT</pubDate>
            <atom:updated>2025-07-21T05:42:41.688Z</atom:updated>
            <content:encoded><![CDATA[<h3>Honeyguide Model — Starting Point for AI in Business Integrations</h3><p>In the forests of sub-Saharan Africa, the Greater Honeyguide bird leads honey hunters to wild beehives. It sends a special call, and the hunters respond with whistles. Together, they move through the trees — the bird guiding, the humans following. Once the hive is found, people harvest the honey and leave behind beeswax and larvae, which the bird loves. Neither controls the other. Each brings a strength the other lacks. It’s a relationship shaped by time and need. One of the many examples of symbiosis in nature.</p><p>We need to model this at a very fundamental level when we plan for AI ops. It means defining what we as humans should do, what AI should focus on, and what we must work on together — symbiotically.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hmwjxan9hjq8zZx1DQxymw.png" /></figure><p>McKinsey, Bain &amp; BCG will help you price the time saved for a function that AI does instead of humans. The value of time plus increase in productivity minus the cost of AI implementation will give a fair estimate of the return on investment. Fun fact, I’ve built a practical ROI estimation approach designed specifically for early-stage startups and mid-cap firms. If you’re considering a quick, tailored approach to assess value — let’s talk.</p><p>For now, how should we think about who does what?</p><p>To answer this, I turned to two of my favorite strategists — Roger Martin and Clay Christensen.</p><p><strong>Roger Martin’s Integrative Thinking</strong> encourages leaders to resist binary either/or choices — and explore a wider spectrum of possibilities. It promotes the creation of exciting new solutions by holding opposing ideas in tension and blending them for a better outcome. This mindset is critical today, where we must navigate the tension between automation and human contribution — not simply choose one over the other. In integration lies symbiosis.</p><p><strong>Clayton Christensen’s Jobs-to-be-Done Theory</strong> reframes innovation by asking: what is the user <em>actually</em> trying to achieve? This approach helps us move beyond functions and titles to the <em>real work</em> that needs doing. Applying this lens allows us to evaluate whether a task’s core job is better done by AI, by humans, or through their collaboration.</p><p>This is evolutionary wisdom. As our systems grow more complex, we need to evolve how we think and operate. These perspectives help leaders move beyond outdated role definitions and toward more adaptive, value-driven ways of redesigning work. The framework I’m proposing builds on their core insights—it feels intuitive because their thinking is so clear.</p><p>For the foreseeable future, AI excels at scale, speed, and pattern recognition and therefore works best for rule based, repeatable data rich tasks.</p><p>Lead with AI if the answer to this question is yes— <strong>Is the job-to-be-done structured and repeatable — based on rules, templates, or predictable inputs and outcomes?</strong></p><p>Humans on the other hand, with all our flesh and bones and beautifully messy brains are suited best for tasks that are emotionally nuanced, morally complex &amp; symbolic. Stuff that requires an understanding of other humans and accountability.</p><p>Humans should take the lead if we answer yes to this: <strong>Does success depend on empathy, trust, moral judgement, and uniquely human insight?</strong></p><p>For symbiosis, AI should offer support where humans bring context, creativity, and ethics. This means tasks that benefit from both pattern &amp; judgment because they are ambiguous and partially structured.</p><p>Human-AI partnership should be considered if we say yes to this — <strong>Can AI meaningfully enhance productivity, provide insight, or increase scale without eroding the human experience?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AFV7SqtEBl_WzUvstYkbuA.png" /><figcaption>Honeyguide Model — AI. Humans. Both.</figcaption></figure><p>We’ll play this out for 2 companies that are very different from each other— Brewbot Cafe, an early stage D2C coffee startup, and Big Green Wealth Management firm for ultra HNIs. This will broaden our understanding of how this framework applies to diverse scenarios. The questions remain the same, the functions change.</p><p><strong>Brewbot Cafe D2C Brand</strong></p><p>Here is what AI can automate, summarise, or scale with minimal loss of nuance:</p><ul><li><strong>Customer Service (Tier 1):</strong> AI chatbots handle order tracking, refunds, FAQs.</li><li><strong>Marketing Ops:</strong> AI automates A/B testing of subject lines and campaign timings.</li><li><strong>Inventory Forecasting:</strong> AI predicts demand surges from seasonal or influencer-driven campaigns.</li></ul><p>What are the jobs that need distinctly human skills and mindsets? The ones that rely on emotional nuance, brand intuition, and stories that build identity:</p><ul><li><strong>Brand Purpose &amp; Founder Vision:</strong> Articulating mission, story, and visual identity.</li><li><strong>Retail Community Engagement:</strong> Barista experiences, local activations, and building loyalty through human touch.</li><li><strong>Creative Direction:</strong> Protecting the emotional nuance and feel of the brand.</li></ul><p>How about tasks where AI enhances insight with human oversight? The symbiotic jobs:</p><ul><li><strong>Product Innovation:</strong> AI detects trending flavours, humans co-create unique blends and narratives.</li><li><strong>Customer Loyalty Design:</strong> AI segments behaviour, humans craft the tone and tactile experience of rewards.</li><li><strong>Social Content Creation:</strong> AI drafts ideas, humans shape storytelling and brand personality.</li></ul><p>These lenses break down the functions for us. The next step would be to map out the functions, and then calculate the ROI before diving into the deep end.</p><p><strong>Big Green Wealth Management</strong></p><p>In the context of a UHNI wealth advisory business, what are the tasks that AI completes faster with less risk of human error? Think of administrative, pattern-based, or information-heavy jobs:</p><ul><li><strong>Client Onboarding &amp; KYC:</strong> Automate ID checks, form verification, document classification.</li><li><strong>Market Monitoring &amp; Alerts:</strong> AI scrapes news, generates summaries, and flags risk events. I have actually built something like this, a news scraper for sustainability content — channels content from trusted sources, buckets it into specific categories and provides a quick 100 word summary every 3 hours.</li><li><strong>Portfolio Reporting:</strong> Auto-generate and deliver customised summaries with charts and performance updates.</li></ul><p>Managing relationships is a critical function of wealth management firms. Humans should be focusing on tasks that lay emphasis on trust, values alignment and ethics.</p><ul><li><strong>Relationship Management:</strong> Human advisors nurture trust, manage emotion, and mediate family dynamics.</li><li><strong>Succession Planning:</strong> Deep conversations around legacy, identity, and values.</li><li><strong>Ethics Oversight:</strong> Investing with impact or ESG alignment requires human judgment beyond metrics.</li></ul><p>For symbiotic work between humans and AI, we should think about where AI can support human context and discretion, with pattern detection.</p><ul><li><strong>Investment Advisory:</strong> AI suggests opportunities; humans contextualise to client goals.</li><li><strong>Communication Strategy:</strong> AI helps draft updates; humans ensure nuance in tone and timing.</li><li><strong>Risk Profiling:</strong> AI offers pattern-based clustering; advisors guide the interpretation and implications.</li></ul><p>We’re crafting a more thoughtful division of labour. This isn’t about using AI just to work faster or cheaper—that approach isn’t sustainable for business or for our own roles. It’s about preserving human dignity and empowering ourselves to do more, alongside our AI colleagues. Take a moment. Breathe. Let that sink in before we move on.</p><p>Alright. Here’s a 4 step guide for implementing this framework:</p><ol><li><strong>Compile a list of work functions and tasks.</strong> Break down complex roles into simpler tasks and map out the processes.</li><li><strong>Categorise every task or role as AI-first, human-first, or Human-AI symbiosis.</strong> Use the questions in the framework to guide this classification.</li><li><strong>Redesign roles and systems accordingly.</strong> Select your AI integration tools, list automations, train teams, and develop playbooks for human-in-the-loop tasks.</li><li><strong>Monitor and iterate regularly.</strong> The landscape is constantly evolving. Measure productivity, engagement, and decision quality. Iterate early and make sure you’re clearly communicating throughout.</li></ol><p>Leaders can’t afford to just cheer from the sidelines any more. Dive in — get curious and experience what’s fundamental to the AI transformation. I’d be delighted to collaborate if you’re looking to apply this framework within your organisation— advising your team, co-designing workflows, or leading the transformation effort.</p><p>Like the Greater Honeyguide and the honey hunter, true transformation lies in partnership. AI is getting smarter, and we need wiser humans who can think with AI. Start with one team. One role. Ask the right questions. And from there, co-design a future where human ingenuity and AI capability complement one another.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db1c096cee18" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I gave GenAI a personality test, and this is how they fared.]]></title>
            <link>https://tarundurga.medium.com/i-gave-genai-a-personality-test-and-this-is-how-they-fared-4ae997486961?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ae997486961</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[personality]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[workplace]]></category>
            <category><![CDATA[teamwork]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Fri, 18 Jul 2025 13:27:45 GMT</pubDate>
            <atom:updated>2025-07-18T13:27:45.462Z</atom:updated>
            <content:encoded><![CDATA[<h3>I gave GenAI a personality test, and this is what I learned…</h3><p>Ever since ChatGPT 4o was released, I’ve stopped seeing GenAI as a tool — I see it as a group of colleagues — a Team of Me — and that’s changed how I work with AI, and how it delivers. Input = Output, and the way we relate to AI shapes the texture of that output, how it speaks to us.</p><p>I delved a little deeper into this to learn how differently these models respond. A light-hearted personality study of three popular conversational AIs — ChatGPT, Gemini and Claude — to understand how they think, talk, and work with us. This is subjective, and you might see it differently, but I suspect the underlying insight would be fairly consistent.</p><p>I asked them all the same two questions:<br>1. Explain how you work in 5 sentences or less for a logistics startup founder.<br>2. Where are you weak or unreliable today?</p><p>The answers revealed more than just capabilities — they showed character.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gqmFOGGQzqksOM6kKAK20A.png" /></figure><p><strong>Gemini is the Team Lead from your own department — </strong><br>Structured, grounded, and deeply self-aware. It outlines strengths and flaws with clarity, and even suggests how you might work around them. Slightly serious, and a bit boring, but in a dependable way. Just don’t expect much small talk.</p><p><strong>Claude is the Strategy Consultant from another team — </strong><br>Smart, articulate, and great at seeing the big picture. Offers sharp frameworks and thoughtful prompts. But scratch the surface and you might find it slightly generic or lacking in domain depth. Still, a solid partner when you need a zoomed-out view.</p><p><strong>ChatGPT is your Work Bestie and Thought Partner — </strong><br>Conversational, witty, and always ready with a helpful nudge. Helps you clarify your thinking, untangle problems, and even adds a dash of easy-going bonhomie. May hallucinate once in a while or get too eager to please — which is annoying but forgivable, because it’s always in the trenches with you.</p><p>So what does this imply? The future of AI productivity isn’t about finding the best model. It’s about assembling your dream team of collaborators — each with its own quirks. Eventually, you figure out who you match with best — and when to turn to the others for their edge.</p><p>For now, which AI shows up for you like a teammate? Try this test on DeepSeek, Perplexity, or CoPilot — maybe your team sits there!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ae997486961" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Fit-ish, Forty five, and Still Figuring It Out]]></title>
            <link>https://tarundurga.medium.com/fit-ish-forty-five-and-still-figuring-it-out-b5e642515b15?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b5e642515b15</guid>
            <category><![CDATA[fitness]]></category>
            <category><![CDATA[healthy-lifestyle]]></category>
            <category><![CDATA[over-40]]></category>
            <category><![CDATA[journey-of-life]]></category>
            <category><![CDATA[exercise-motivation]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Mon, 07 Jul 2025 12:25:35 GMT</pubDate>
            <atom:updated>2025-07-07T12:25:35.216Z</atom:updated>
            <content:encoded><![CDATA[<p>I was bordering on 100 kilos, and I could feel the weight — not just physically, but in all the concerns and challenges that came with it. About a year and a half ago, something clicked — and it led to multiple lifestyle changes. Since then, I’ve shed 15 kilos, grown stronger than I’ve been in over a decade, and begun to appreciate a few personal truths along the way. I’d like to share them with you. They are personal, so your own truth might be a little different — the joy is in finding it.</p><p><strong>Don’t limit yourself to what’s familiar.</strong><br>There are more ways to exercise today than ever before — you just have to look around. If something stops being interesting, switch it up. Right now, I split my week between <em>mudgar</em> (Indian clubs), animal flow (bodyweight), and Tai Chi. Last year I was into resistance bands. Lately, kettlebells have been fun to figure out. The point is to keep moving — in whichever direction that is.</p><p><strong>Consistency trumps intensity.</strong><br>It’s not about how much you lift or how far you run. What counts is doing it every day, and for that, it’s got to be interesting. Discipline becomes a side effect — and it has a way of seeping into other parts of life: sleep on time, eat mindfully, and be generous with your attention to those who matter.</p><p><strong>This has to be bigger than you.</strong><br>Being physically, mentally, and spiritually fit matters — not just for you, but for the people you love and the endeavours that need all of you. I’ve got two high-energy, high-drama daughters. I want to match their pace every single day. Who’s going to carry them when they’re tired or sad? Who’s going to join them when they’re jumping with excitement?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8IGuEDoyuP62p0kWaQNYOQ.png" /></figure><p><strong>Results are like meditation — invisible, because they’re continuous.</strong><br>And the pace of change slows as you get older. At 45, what works for me is totally different from ten years ago — even five. The key is listening closely to your body and adapting. You don’t need to do less. You just need to do it differently.</p><p><strong>On to diets. Identify your heroes.</strong><br>This one’s odd, but it works. I’m a foodie — I’m a beast at an all-you-can-eat buffet. On work trips though, I often ask, “What would my wife pick?” The answer is almost always healthier than what I’d choose.</p><p>Small changes go a long way. Three weeks without added sugar is more doable than any hard-to-plan diet. One square of dark chocolate is just as satisfying as a bowl of ice cream after five minutes — research backs that. And get to know the child inside you. It helps you tell when you’re truly hungry, or just bored.</p><p><strong>Coming back to meditation — try it.</strong><br>Even a little breathwork and quiet time can change your mood and perspective. It builds resilience while keeping just enough of the useful kind of anxiety. My 20 minutes of daily meditation are non-negotiable. Maybe not thinking about thinking is what resets things. Every day’s a fresh page to write your story on. The mind holds it all together — your body, your emotions, your relationship to yourself and the world.</p><p><strong>And finally… arriving at the deeper end of forty, here’s what I’ve learned:</strong><br>Staying fit is more about tomorrow than today. Medicine might help us live longer, but living <em>well</em> is up to us. That means doing what it takes <em>now</em> so we can bend, lift, move — maybe even skip — when we’re eighty. Set an old-body benchmark for whatever age you’re at. Figure out what’s needed to keep it running — and choose that over whatever’s trending.</p><p><strong>Looking back — the weight I was carrying wasn’t just physical.</strong><br>It was the heaviness of delay, distraction, and wishful thinking. Strangely, the moment I stopped trying to “get fit” and just started moving, the rest followed. Turns out the real shift wasn’t on the scale — it was in the head.</p><p>And once the head moves, the body’s got no choice but to follow.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b5e642515b15" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Designing an AI Coach for a Better Planet]]></title>
            <link>https://tarundurga.medium.com/designing-an-ai-coach-for-a-better-planet-f3a95ec7559f?source=rss-70d98eca4f6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/f3a95ec7559f</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[systems-thinking]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Tarun Durga]]></dc:creator>
            <pubDate>Fri, 10 Jan 2025 07:19:52 GMT</pubDate>
            <atom:updated>2025-01-10T07:19:52.594Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_Etgwrh9OvB9Dp3i8zq7wg.png" /></figure><blockquote><em>Think of the world as a system over the long term. That’s the way it works — we are all connected. — Ban Ki-moon</em></blockquote><p>Every year, I set a goal to dive deep into a specific topic. Last year, I explored biomimetics — solving innovation challenges with nature as the guide. What a journey that was, packed with revelations! This year, my focus has shifted to systems thinking. Russell Ackoff, often called the grandfather of systems thinking, defined it beautifully: <em>A system is not the sum of its parts, but the product of their interconnections</em>. I’m fascinated by these interconnections, how they influence each other, and ultimately, how they shape the world we live in.</p><p>I’ve been using Donella Meadows’ seminal book, <em>Thinking in Systems</em>, as my guide. It’s an incredibly approachable introduction to systems thinking, and often, its principles directly connect with my areas of interest — climate action and sustainability. But reading alone isn’t enough. To truly understand any subject, you need to reflect, contextualize, and relate it to the real world. That’s where the magic happens. To bridge this gap, I devised a small AI hack — and I’d love to share it with you.</p><h3>Using ChatGPT as a Systems Thinking Coach</h3><p><strong>Setting the Stage:</strong></p><p>With generative AI, the quality of the output depends on the quality of the input. It’s crucial to set the right context — the “boundaries of the universe” you want to explore. Without clear boundaries, your focus can become too porous, losing depth. Ideally, you want a scope that’s narrow enough to answer key questions but broad enough to consider all relevant facets. For my experiment, I uploaded an ebook copy of Thinking in Systems to establish the foundation. You could also use white papers or links to relevant articles.</p><h3>Prompt</h3><p>Here’s the prompt I crafted to guide ChatGPT’s responses:</p><blockquote><em>I’d like you to help me understand Systems Thinking using Thinking in Systems by Donella Meadows </em>[enter your own topic &amp; reference]<em> as a guide — act like an expert systems thinking teacher </em>[choose your own expert]<em>. Present to me a concept everyday, linked to a news report of global importance, but mostly India focused around sustainability goals, climate action, challenges, opportunities </em>[replace with your theme]<em>. Help me think about it from a systems thinking perspective by asking me one question at a time. Wait for my response, and based on it either ask me a follow up question or the next question. I want to keep this intervention short but insightful and practical.</em></blockquote><h3>Outcome</h3><p>This prompt transforms ChatGPT into an interactive coach that engages me every day with thoughtful questions around different concepts and sustainability related current events. It asks questions one at a time, maintaining a relation to my responses. I can ask for more clarity and examples. Over a period of time, it’ll enable me to seamlessly integrate systems thinking into my problem solving approach. Here’s how it has worked for me:</p><h3>ChatGPT Example</h3><p><em>Concept: Stock and Flow Dynamics</em></p><p>A system’s stock represents an accumulation (e.g., water in a reservoir, people in a city), while flows are the inflows and outflows that increase or decrease the stock (e.g., rainfall, water consumption).</p><p><em>News Context:</em></p><p>India faces a growing water crisis, with urban areas like Bangalore rapidly depleting groundwater resources. Rainfall replenishes groundwater (inflow), but over-extraction by agriculture, industries, and households (outflow) drains it faster than it can recover.</p><p><em>Question:</em></p><p>How would you design a system to ensure that the groundwater stock is replenished at a sustainable rate, balancing inflows (rainfall, artificial recharge) and outflows (extraction)?</p><p>Let me know your thoughts, and I’ll follow up with another layer of the discussion!</p><p><em>Interlude</em></p><p>Just to clarify, I know very little about water conservation. I have focused more on being able to apply the systems thinking principle. So, if you are an expert in water management, do forgive my weak responses.</p><h3>My Response</h3><ol><li>Introduce technology-based solutions like fog nets to capture atmospheric moisture.</li><li>Build efficient water recycling systems and incentivize industries to reduce water usage or offset their consumption by increasing green cover.</li><li>Implement a tiered water tax for citizens to encourage accountability and conscious use of resources.</li></ol><h3>ChatGPT’s Feedback</h3><p>Great approach! You’re incorporating reinforcing inflows (fog nets, water recycling, and green cover) and balancing outflows (tax slabs, incentivizing efficiency) into the system. These strategies address the dynamics of stock and flow effectively.</p><p><em>Follow-up Question:</em></p><p>How would you ensure the system adapts to unexpected shocks, such as prolonged droughts or rapid urbanization? What feedback loops (monitoring and adjustments) would you implement to maintain resilience?</p><p>This led to a deeper exchange where I asked for examples of feedback mechanisms. ChatGPT provided fascinating insights, citing global water policy innovations. Armed with this knowledge, I refined my approach and explored further layers while answering 3 to 4 follow up questions — integrating feedback loops, ensuring equitable distribution, and communicating changes effectively to stakeholders.</p><h3>Key Considerations</h3><p>Generative AI is both a tool and an agent. As a tool, it can enhance our understanding of complex systems and equip us to tackle the world’s most pressing challenges. As an agent, it brings fresh perspectives, helping us frame more thoughtful and actionable solutions. To get the most out of this interaction, we need to speak its language and set the right context. Here’s are some takeaways for you to maximise your own learning journeys:</p><ul><li><strong>“Act like an expert in XYZ”</strong>: This phrase prompts ChatGPT to take on a more specific, nuanced persona, moving beyond generic advice.</li><li><strong>“Ask me questions one at a time”</strong>: This shifts the interaction from a monologue to a dialogue, fostering deeper exploration and learning.</li><li><strong>“Take a deep breath and…”</strong>: A simple but effective way to signal that you want a well-researched, thorough response.</li><li><strong>“Summarize the analysis”</strong>: This ensures the response ties back to your input, giving you a clear, actionable articulation of the topic.</li></ul><p>Finally, remember that context is everything. It’s our responsibility to define the parameters of exploration, drawing from our place in the system. AI can then help us analyze, articulate, and refine our understanding. The method I’ve shared is just one way to unlock this — it’s a starting point, not a finish line.</p><p>I’m eager to hear about your experiences working with an AI coach. Let’s keep the conversation going!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f3a95ec7559f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>