<?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 Michael Akinwumi on Medium]]></title>
        <description><![CDATA[Stories by Michael Akinwumi on Medium]]></description>
        <link>https://medium.com/@datawumi?source=rss-4b97e98bc72c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*inQXumeidwaH5JCLjXf39A@2x.jpeg</url>
            <title>Stories by Michael Akinwumi on Medium</title>
            <link>https://medium.com/@datawumi?source=rss-4b97e98bc72c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 19:14:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@datawumi/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 Ecosystem Has No Referee]]></title>
            <link>https://medium.com/@datawumi/the-ecosystem-has-no-referee-5e682714c135?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/5e682714c135</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[workforce-development]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Sat, 24 Oct 2026 21:38:11 GMT</pubDate>
            <atom:updated>2026-10-24T22:00:23.585Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Microsoft’s CEO names a real AI risk. His fix has no regulator, no seat for workers, and no one who can say no.</em></p><p>Last week I argued that the skill that will matter most in the AI age is deciding what to trust. Machines now produce knowledge faster than people can read it. So, someone must evaluate, verify, and validate the output, or we end up <a href="https://datawumi.substack.com/p/take-no-machines-output-for-it">taking every machine’s output for it</a>. This time I reconsider that skill seriously and hands it to the company.</p><p>This series, introduced in <a href="https://substack.com/home/post/p-213457294">Governing the Governors</a>, holds public statements from AI leaders to four questions. What harms does a statement name, and which does it avoid? Who is accountable, and through what mechanism? What risk is it actually addressing, and is the fix proportionate? Whose interests does its structure serve? The thesis is old arithmetic:</p><p>“Those in power are few. The people are many. Every fallen tyranny eventually discovered the arithmetic.”</p><p>The last reading, <a href="https://datawumi.substack.com/p/below-the-threshold">Below the Threshold</a>, looked at a lab’s proposal for binding rules on the largest models. This one looks at the other side of the market: what a platform company says businesses should do about those models.</p><p>In June, Microsoft’s chairman and CEO posted <a href="https://x.com/satyanadella/status/2066182223213293753">a 15-paragraph essay on X</a> arguing that a frontier without an ecosystem is not stable. The diagnosis is sharp. Frontier models can absorb what a company knows and resell it at commodity prices. If a handful of models capture the returns, he writes, the public will not tolerate it. His remedy is for every company to build its own learning loop on top of the models. The loop turns the firm’s workflows and its people’s judgment — collectively called human capital — into what he calls token capital. Private evals measure it, internal work traces train it, and it is agnostic to model vendors.</p><p>It is the right problem. Here is how his statement ranks against the four questions.</p><h3>1. The harm list is written from the vendor’s desk</h3><p><a href="https://doi.org/10.1016/j.patter.2026.101517">MIT’s AI Risk Repository</a> is a catalog built from 74 published AI risk frameworks, and it sorts AI harms into 24 types. The essay names three of them, all economic: power concentrating in a few systems, displacement across industries, and the devaluation of human expertise. The frameworks MIT reviewed cover eight on average. An essay is not a risk framework, but the gap is still telling.</p><p>There is nothing on discrimination, privacy, safety or misuse. There is also nothing on energy, from a company whose <a href="https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/2026-Microsoft-Environmental-Sustainability-Report-PDF.pdf">own sustainability report</a> says its total emissions rose 25% year over year in 2025, driven mainly by datacenter expansion and pause on renewable energy use.</p><p>The victims in the essay are companies, industries, and countries. People appear in only two roles: as the source of the expertise to be captured, and as employees whose judgment becomes part of the firm’s systems. The tenant, the borrower, and the job applicant never appear, though they are the people AI decisions are made about.</p><p>Throughout the essay, it is the <em>models</em> that absorb and capture value. The companies that build, price, and sell those models are never called to act.</p><p>A list of harms tells you whose desk it was written at.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aTZl1QGL2H60CxxHw5pX6g.png" /></figure><h3>2. Accountability: everyone, so no one</h3><p>Who is asked to act? Enterprises, by building their own loops. Platforms, by following the author’s personal belief that a platform should create more value for those who build on it than it keeps for itself. That is the whole list. In 15 paragraphs, no regulator, legislature, court, or public agency appears. The only outside check is a forecast that the political economy will not tolerate concentration. The essay closes with a call to build a stable equilibrium together, without saying who is meant to build it.</p><p><a href="https://arxiv.org/abs/2306.06924">TASRA</a> is a taxonomy that sorts large-scale AI risks by who is accountable for them. It has a category for exactly this shape: <em>diffusion of responsibility</em>, where harm emerges from many actors’ choices and no one in particular answers for it. Its verdict on that category is blunt: “More than individual company audits will be necessary.” It calls for regulating algorithms the way we regulate food and drugs, and for institutions that can watch the algorithmic economy as a whole.</p><p>The essay prescribes the reverse: every firm protects itself. But the harm it describes is not diffuse. A few dominant models means a few model makers. A harm with a short list of authors has been filed under a category with none.</p><p>When everyone is responsible for a market, no one is accountable for it.</p><p>There is a later statement — I intend to cover in a future episode — worth crediting. Four months on, the same executive <a href="https://x.com/satyanadella/status/2108931348857827686">wrote</a> that “[v]alidation must be independent of the intelligence being validated.” That is a stronger principle than anything in the June essay. But in both essays, independence means independence from the <em>model</em>, not from the <em>firm</em>. The one checking the work is still the house and, as they say, the house never loses.</p><h3>3. Right risk, wrong level</h3><p>The mechanism the essay identifies is real: models soaking up expertise and concentrating its returns. The fix is aimed at the wrong level. Concentration is a property of markets, and markets have tools for it: competition law, interoperability standards, and data-portability rights. Better internal systems at individual firms do not change who holds pricing power at the model layer.</p><p>The essay’s brilliant idea is being model agnostic, that a firm should be able to swap one model for another without losing what it has built. It calls this a test of sovereignty, then leaves it voluntary. There is no standard, no requirement, and no one to enforce it.</p><p>The remedy also has a price of admission. Private evals and private training environments take data, engineers, and capital. Every company is invited, but large incumbents are the ones who can afford to attend. My inference is that the firms most exposed to commoditization are the least able to build this defense.</p><p>And the remedy moves concentration down a floor instead of removing it. The essay warns against a few models capturing the value, then praises the firm-level loop because it compounds into an advantage rivals cannot easily copy. Concentration is a problem at the model layer and a strategy one layer down.</p><p>Then there are the evals. The essay says private evals should test whether a model is improving on outcomes that matter to the business. TASRA warns about this kind of instrument: a system tuned on the signal its owner can measure while its effect on people goes unmeasured. A claims model scored on processing cost will look excellent. Whether the claims it denied were fair is not on the scorecard.</p><p>An eval scored only by the business will find what the business is looking for.</p><p>The essay’s most honest line concerns the last time an economy went through this. During globalization, it says, “[t]he GDP numbers looked fine on the surface, but the displacement was real,” and the consequences are still being felt. Economists <a href="https://www.nber.org/digest/dec14/import-competition-and-great-us-employment-sag">estimate</a> that Chinese import competition cost the United States a net 2 million to 2.4 million jobs between 1999 and 2011. That displacement fell on workers and towns, not on the firms that moved production. The essay’s remedy protects firms. Inside each firm, it turns employee judgment into a company asset that can be replicated and scaled. And training on internal work traces means training on records of employees’ daily work, with consent never mentioned. The same pattern repeats at a smaller scale: the firm’s numbers can look fine while the people whose expertise trained the loop absorb the cost.</p><p>To understand the coming layoff realities if risk assessment focuses on the wrong level, Barclays has said that it will <a href="https://www.reuters.com/business/finance/barclays-q3-profit-rises-12-announces-670-million-buyback-2026-10-22/">eliminate 5,000 jobs this year due to AI</a>, Uber has doubled down on <a href="https://www.businessinsider.com/uber-uses-openais-chatgpt-work-after-major-layoffs-2026-10">AI for workers after recent layoffs</a>, and Sundar Pichai claims <a href="https://www.businessinsider.com/sundar-pichai-explains-how-ai-agents-help-him-run-google-2026-10">a team of AI agents helps him run Google</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*agNbR0ji6za8c_UVgPkXcg.png" /></figure><h3>4. Whose ecosystem?</h3><p>If models become interchangeable parts, value moves to whoever sells the shelf they sit on. Microsoft’s Foundry catalog lists <a href="https://learn.microsoft.com/en-us/azure/foundry/concepts/foundry-models-overview">more than 10,000 models</a>, from providers including OpenAI, Anthropic, Meta, and Mistral. A world where firms swap models freely is good business for whoever hosts the swap. The essay never names Microsoft. This reading is mine, not the essay’s.</p><p>Large enterprises gain a playbook and a moat. Countries are invoked in the language of sovereignty. Smaller firms are invited without the means to take part. Workers supply the judgment and get no stated share, no consent, and no portability of their own. The essay insists that a firm must keep its expertise when it changes models. It says nothing about an employee keeping theirs when they change jobs or when they are laid off due to AI-driven efficiency.</p><p>Portability is offered to firms, not to people.</p><p><a href="https://arxiv.org/abs/2507.09369">A Taxonomy of Omnicidal Futures</a> sketches how AI could go catastrophically wrong without anyone intending it. It follows this economic logic to its far end, with human employees reduced to a “physically embodied rubber stamp” for decisions made by machines. The essay is nowhere near that world. But its assurance that humans will keep directing the loop is stated as a belief, and nothing in the proposed architecture secures it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s0Z-3DUjYD8hrGRmJUQ_9g.png" /></figure><h3>In fairness</h3><p>This essay does things most industry statements do not. It names concentration of economic power as a harm, and it comes from the head of a company well placed to benefit from that concentration. It concedes that AI needs society’s permission to proceed. It puts a concrete idea against lock-in on the table, in model portability. And its line about globalization names, plainly, the gap between aggregate numbers and lived consequences that this series exists to measure.</p><h3>The house and the many</h3><p>Last week’s question was who evaluates the machine. This essay answers it: the firm, against its own business outcomes, in private. Evaluation, verification, and validation become one more asset on the balance sheet.</p><p>The AI developers are few. The deployers are more. The people whose judgment trains the loop are many. Until the many can read the scorecard, the house is still grading itself.</p><h3>What is Coming</h3><p><em>Expect another statement, the same four questions. Coming from an inspiration from an introductory class on decision trees — covering Iterative Dichotomizer 3 (ID3), boosting and bagging algorithms — I taught my students this week, I plan to use next week to focus on how representation influences algorithmic decisions. I will then return to the Governing the Governors series by applying the standard four-question analysis to Demis Hassabis’s (Google DeepMind) framework for frontier AI.</em></p><p><em>In his July public statement published, he argues that artificial general intelligence is “probably only a few short years away” and proposes a concrete U.S.-led governance mechanism: a “Frontier AI Standards Body,” modeled on the Financial Industry Regulatory Authority (FINRA), that would independently test frontier-class models for cybersecurity, biological, and other national-security risks before release. Under the proposal, participation would begin voluntarily, with frontier labs sharing models up to 30 days ahead of release, before potentially becoming a mandatory condition of U.S. market deployment. Hassabis frames the U.S. moving first as a starting point for eventual international coordination on AI safety standards.</em></p><p><em>I look forward to sharing the outcomes of my four-question X-ray of the statement.</em></p><p><em>Dr. Michael Akinwumi is Chief AI Officer and Head of the Responsible AI Lab at the National Fair Housing Alliance (NFHA) and an adjunct professorial lecturer in decision sciences at the George Washington School of Business (GW Business). The views expressed are the author’s alone and do not reflect the views, positions, or policies of NFHA or GW Business.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5e682714c135" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[In 2276]]></title>
            <link>https://medium.com/@datawumi/in-2276-326dc8c9239c?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/326dc8c9239c</guid>
            <category><![CDATA[history]]></category>
            <category><![CDATA[american-history]]></category>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Sat, 04 Jul 2026 02:31:22 GMT</pubDate>
            <atom:updated>2026-07-04T02:31:22.474Z</atom:updated>
            <content:encoded><![CDATA[<p>America, look at you at two hundred and fifty.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tMy4NeyVINKmBBNS5aRjiQ.jpeg" /><figcaption>America at 250 cake</figcaption></figure><p>I came to know you first the way the world does — from the outside, through the news and the many groundbreaking things that came out of you. But that was never what drew me. What drew me was an idea — the improbable claim, written down before it was ever true, that a person’s imagination and willingness to work, and not the accident of birth or bloodline, should determine what she becomes. My moral philosophy was shaped by the work and the lives, the aspirations and the sufferings of your sons and daughters — read about first, and then lived among. I became one of your residents in the last years of your first two hundred and fifty. From outside you are an idea, luminous and simple; from inside you are an argument, with yourself, never quite settled. I will not be here for your next two hundred and fifty years. But I count it a privilege to write to those who will be.</p><p>You are, and have always been, an experiment. You were born in 1776 out of borrowed Enlightenment light — natural rights, the rule of law, government by consent — that no nation had yet dared to build a country upon. Yet an experiment is not a proof, and from the first day your hypothesis was set against its own contradiction. Many of the men who signed the Declaration of Independence, declaring it self-evident that all men are created equal, owned other men even as they signed. Roughly one in five people in the thirteen colonies was enslaved, and eleven years after the Declaration of Independence your framers wrote protections for slavery into the Constitution itself. This is your peculiar genius and your peculiar torment: you gave your children a promise larger than your practice, and then had to live for centuries with the distance between them.</p><p>The temptation, when that distance grows unbearable, is to postpone the reckoning rather than face it. In 1820, rather than resolve whether a nation half-slave could remain a nation at all, you drew a line across the map, with slavery permitted below and banned above, and called it a compromise. It bought you forty years of quiet and cost you six hundred thousand dead. A reckoning deferred is not a reckoning avoided; it only accrues interest, paid in a harder currency than the principal. I write this because you are being tempted again to draw new lines and call them peace.</p><p>Let me tell you the part of your story that is also my own. You have always been renewed by the people who chose you rather than those who merely inherited you; the immigrant does not take you for granted, because you were never guaranteed to her. During your Civil War you threw your doors open — the Homestead Act, the plainly named Act to Encourage Immigration — and by 1910 nearly one in seven of your people had been born somewhere else. And yet every time your economy shook, you turned on the people you had invited. In 1882 the Chinese Exclusion Act closed your doors to Chinese workers; four years later you dedicated a statue in your harbor whose meaning the whole world took to be the welcome of all comers. Every generation of newcomers has forced you to answer the same question: who does the promise belong to? Your best answer has always been the same — anyone willing to work for it, regardless of where they came from.</p><p>There is an older answer beneath that civic one. At the moral root of much of your culture is a teaching in which a people is measured not by how it treats the powerful but by how it treats the stranger: I was hungry, and you fed me; I was poor, and you clothed me; I was without shelter, and you took me in. The stranger at your door is not a threat to the promise. The stranger is the test of whether the promise was ever real. I am one of your strangers. You took me in, and I have tried to be worthy of it — which is why an immigration system worthy of your future would be open to imagination, rewarding of effort, and indifferent to birthplace. I am myself a small piece of evidence it works.</p><p>Now to the most hopeful thing in your recent history. For most of a century after Reconstruction, legal segregation ruled your South, and argument had not moved your citizens. A preacher named Dr. Martin Luther King led disciplined, nonviolent demonstrations, and for years most white Americans regarded him as divisive, even after he told the country of his dream. What finally moved them was a machine. In the spring of 1963, in Birmingham, Alabama, a police commissioner turned high-pressure fire hoses and dogs on black children marching for their freedom — and the cameras were there, and the footage went into living rooms across the country. Suffering that had been invisible because it happened elsewhere was suddenly visible at scale. The television camera was not moral. But it made the invisible visible, and a nation that could no longer claim not to know was shamed into acting. Within a little over a year came the Civil Rights Act of 1964; the next year, the Voting Rights Act. Technology, at its best, does not make us good. It makes it harder to pretend we do not see.</p><p>But do not mistake me for a person who believes machines redeem us. The very next transformative technology proved the opposite. You built the internet, and then companies whose algorithms learned to serve people more of whatever they craved most — outrage, conspiracy, the strident voice over the reasoned one. It did not make suffering visible so much as make anger profitable, and it left you more divided than it found you. The difference between the camera in Birmingham and the algorithm in your children’s pockets is not sophistication. The difference is governance. One entered a society with institutions capable of turning what people saw into what a nation did. The other entered largely ungoverned. Technology is not destiny; governance is.</p><p>Artificial intelligence has now entered your mainstream, with sums poured into it that reorder the plans of your largest companies. It has raised the oldest question your history keeps asking: who should control a transformative power — the public, through its institutions, or a handful of private hands rich enough to build it? To make this new intelligence an instrument of repair rather than a repetition of the social-media wound, you will need the unglamorous machinery of accountability: a right to be told why a machine decided as it did about your loan, your job, your liberty; a right to contest it before a human who can overturn it if found culpable; real liability when these systems cause harm; and independent auditing that makes patterns of harm invisible to any single victim visible at scale. Built with those safeguards, it could make the gap between your promises and your practices impossible to ignore. Built without them, it will amplify division and leave the most vulnerable more exposed than before. The technology will not choose. You will.</p><p>I must tell you plainly what I see now. The line between public office and private interest is blurring again. Loyalists are replacing independent public officials. The norms and institutional checks that let a democracy correct itself are eroding, and your polarization has reached a point where too many people can no longer disagree without wishing to destroy one another. This is the faltering. But I know you have faltered before — this is not even your worst hour, not by the measure of 1857 or 1861 or the long betrayal after Reconstruction — and I know you have recovered before.</p><p>There is a line in one of your oldest inherited scriptures that has become, for me, the hinge of this letter: “If the foundations be destroyed, what can the righteous do?” It is usually read as a cry of despair. But I have come to read it as a question that demands an answer rather than a lament that forecloses one. What can the righteous do, when the foundations are shaken? The answer is the plainest thing in the world. They can pick up the stones. They can rebuild — one repaired institution, one honest official, one governed machine, one welcomed stranger, one generation at a time.</p><p>You have heard this call before. In the dark winter of the Revolution, Thomas Paine wrote that “these are the times that try men’s souls,” and warned that “what we obtain too cheap, we esteem too lightly.” You are in need of another dose of Paine’s resolve. The centuries ahead will ask you to pay again — not, I pray, in the currency of 1861, but in the daily coin of vigilance: of defending your democratic infrastructure, your elections and your institutions, against those who would corrupt it, whether by cyber attack from abroad or by corruption within. That is your oldest duty in modern dress: to protect the conditions under which a free people can govern itself.</p><p>Let me tell you what I see when I imagine you in 2276, if you choose wisely now. I see a republic in which the line between public office and private interest has been redrawn and, this time, held. I see a country that has finally reckoned with the black cloud of slavery and its long aftermath — not through the endless circulation of guilt, but through the kind of repair that leaves the foundation stronger than it was before the crack. I see an immigration system open to imagination, rewarding of effort, indifferent to birthplace. I see a democracy that governs transformative technology so that its benefits flow broadly and its harms fall on no one without recourse. And I see a country that remains the beacon of the free world — not because it is perfect, but because it has never stopped trying to be, having learned the hardest democratic skill of all: how to disagree without destroying.</p><p>America, I am nearly done. I will not see 2276; I will be, by then, one of the countless residents who loved you in their turn and passed on. A woman does not plant a slow-growing tree because she expects to sit in its shade. From outside, I believed you were the one place on earth where a person’s future was not decided by her past. From inside, I have learned what the idea did not prepare me for: that you are not a finished thing that keeps this promise automatically, but an unfinished thing that keeps it only when your people choose, again and again, to keep it. You are not a guarantee. You are a labor.</p><p>When your Constitution was finished, the story goes, Benjamin Franklin was asked what kind of government you had been given, and he answered: “a republic, if you can keep it.” For two and a half centuries those last five words have been read as a warning, and they are one. But I want to hand them to the future as an act of faith instead. A republic, if you can keep it: I believe you can keep it — not because I have found you flawless, but because I have seen what you are made of, and I have seen who keeps arriving at your door, still believing, as I did, that it is worth the work. I staked my life on that belief when I chose you, and I stake it again now, to the people I will never meet.</p><p>Keep it.</p><p><em>With love, and in hope,</em></p><p><em>Michael Akinwumi,</em></p><p><em>one of your residents,</em></p><p><em>in the final years of your first two hundred and fifty.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=326dc8c9239c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Code, Community, and Environment: Key Tech and AI Takeaways from the NFHA 2026 National Conference]]></title>
            <link>https://medium.com/@datawumi/code-community-and-environment-c864673e0a20?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/c864673e0a20</guid>
            <category><![CDATA[data-center]]></category>
            <category><![CDATA[housing]]></category>
            <category><![CDATA[ai-policy]]></category>
            <category><![CDATA[fair-housing]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Fri, 03 Jul 2026 03:48:56 GMT</pubDate>
            <atom:updated>2026-07-03T03:49:44.566Z</atom:updated>
            <content:encoded><![CDATA[<p>The National Fair Housing Alliance (NFHA) annual conference brought together a powerful coalition of advocates, legal experts, and community leaders to confront a rapidly changing housing landscape. Amidst a challenging federal environment marked by a pullback in traditional civil rights enforcement, the conference focused on developing bold, localized solutions to protect vulnerable households. Discussions throughout the event made it clear that achieving housing justice today requires looking far beyond physical real estate offices and traditional landlord-tenant disputes. Instead, the fight has moved to the cutting edge of technology, where digital tools and infrastructure are reshaping neighborhoods and community health alike.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sRfHGhcyFlvz6PjASfgGbg.jpeg" /></figure><p>A primary focus of the conference was the massive, unchecked footprint of artificial intelligence and big data on the housing sector. In an interactive workshop, facilitators demonstrated how commercial AI systems — increasingly used by over 82% of Americans for housing insights — reproduce historical patterns of discrimination. For instance, testing of real-time AI search models revealed blatant racial steering: when a white home seeker profile queried an AI model about New Orleans, the system steered the user exclusively to affluent, predominantly white neighborhoods while completely omitting diverse, historically Black neighborhoods of equal character. Furthermore, the AI models disproportionately associated terms like “diversity” and “progressiveness” only with Black search profiles, while leaving white search results completely devoid of these descriptors. This is consistent with findings documented in a recent study completed by NFHA’s Responsible AI Lab. A preprint of the study is <a href="https://arxiv.org/abs/2606.06694">here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*06_Pg58wt8Y5HlbxoGIF7Q.jpeg" /></figure><p>In a separate panel discussion, we learned that the digitization of housing search tools is closely mirrored by the rise of under-regulated tenant screening algorithms. Panelists warned of an “algorithmic monoculture” where landlords increasingly outsource housing decisions to a handful of predictive tech models. These black-box systems generate speculative risk scores using biased, historical eviction and credit data to predict future renter behavior, effectively locking out thousands of qualified applicants. To counter this, housing advocates are pushing for fairer screening alternatives, such as positive-only rent reporting and cash-flow underwriting under the protection of Section 1033 of the Dodd-Frank Act, while continuing to utilize the Fair Housing Act’s disparate impact theory to challenge discriminatory automated models.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S0BkHvW218UqOq1jOOs_GQ.jpeg" /></figure><p>The conference also drew critical connections between the virtual world of AI and its physical consequences, dedicating significant focus to the environmental and housing toll of hyperscale AI data centers. These massive data facilities are being built at an unprecedented velocity, frequently adjacent to low-wealth communities of color. A prominent case study from Memphis illustrated the severe cost of this infrastructure: dozens of methane gas turbines without Clean Air Act reviews were allegedly deployed, leading to spikes in pediatric respiratory emergency room visits in a historic Black neighborhood and causing families to leave the area entirely.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ricfYvUg53WAabg99DGdcg.jpeg" /></figure><p>To reclaim local control, community leaders are using municipal zoning enforcement to address noise and air pollution, while pointing to resources like the NAACP’s national data center principles and templates for legally binding Community Benefit Agreements (CBAs) to prevent big tech from exploiting local resources at the expense of ratepayers. The conference emphasized that as technology continues to scale, local organizing and strict data accountability must serve as our primary line of defense.</p><p>Ultimately, the major takeaway from this year’s conference is that the virtual and physical footprints of technology can no longer be evaluated in isolation. Whether we are fighting the invisible steering of an online search algorithm or the very real pollution of a local data center, our defense remains rooted in community-led action. As the federal regulatory landscape shifts, the responsibility to establish guardrails falls to local advocates, municipal planners, and state coalitions. By uniting to demand transparency from developers, enforcing local zoning protections, and centering human dignity over speculative tech growth, we can ensure that the technological advancements of tomorrow do not come at the expense of our neighborhoods today.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c864673e0a20" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The First Miracle Wasn’t About Wine. It Was About Living Well.]]></title>
            <link>https://medium.com/@datawumi/the-first-miracle-wasnt-about-wine-it-was-about-living-well-96a19caed27e?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/96a19caed27e</guid>
            <category><![CDATA[motivation]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[lifestyle]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Sun, 28 Jun 2026 02:58:44 GMT</pubDate>
            <atom:updated>2026-06-28T02:58:44.702Z</atom:updated>
            <content:encoded><![CDATA[<p>The first recorded miracle of Jesus in John 2:1–11 (GNTD) is often remembered as the moment water became wine. Yet beneath the miracle lies a series of lessons about human development principles including purpose, relationships, leadership, and personal transformation.</p><h3>1. Great Purpose Doesn’t Isolate You from Ordinary Life</h3><p>The story begins not in a synagogue or palace but at a wedding.</p><p>Jesus and his disciples accepted an invitation to celebrate with others.</p><p>That is instructive.</p><blockquote>Purpose is not an excuse to withdraw from society.</blockquote><p>The most influential people know how to participate in ordinary human experiences. They celebrate with others, build relationships, and remain present in community.</p><p>Sometimes the next opportunity appears not in isolation but around a dinner table.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/543/1*6VuyPmlfjmhctr8TREMW7Q@2x.jpeg" /></figure><h3>2. Emotional Intelligence Often Precedes Miracles</h3><p>Mary notices a problem before anyone publicly acknowledges it.</p><blockquote>They have no more wine.</blockquote><p>She doesn’t panic.</p><p>She doesn’t embarrass the hosts.</p><p>She quietly communicates the need.</p><p>This is emotional intelligence in action.</p><blockquote>She observes.</blockquote><blockquote>She discerns.</blockquote><blockquote>She responds with wisdom.</blockquote><p>Many opportunities are lost not because people lack intelligence, but because they lack sensitivity to people and timing.</p><h3>3. Transformation Begins When We Question Old Containers</h3><p>Jesus instructs the servants to fill ceremonial purification jars with water.</p><p>These vessels represented an existing religious tradition.</p><p>Instead of introducing something entirely new, Jesus transforms what already exists.</p><p>Innovation often works this way.</p><blockquote>The greatest breakthroughs are frequently born by reimagining familiar structures.</blockquote><p>Purpose is rarely about discarding everything.</p><p>It is about redeeming what can become something greater.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/543/1*L0SV5osoA4kjrEt_yEkXRA@2x.jpeg" /></figure><h3>4. Your Calling Is Bigger Than Who Gets to Witness It</h3><p>One subtle feature of the Cana story is who is present and who is absent.</p><p>Mary is there.</p><p>The disciples are there.</p><p>The servants are there.</p><p>Joseph is not mentioned.</p><p>Whether because he had died or for another reason, the narrative reminds us of a timeless reality:</p><blockquote>Not everyone who helped shape your beginning will witness your greatest moments.</blockquote><p>That is not tragedy.</p><p>It is simply the nature of life’s unfolding.</p><p>Purpose unfolds across generations.</p><blockquote>Some plant.</blockquote><blockquote>Others water.</blockquote><blockquote>Others witness the harvest.</blockquote><h3>5. Your Glory Exists to Reveal, Not Impress</h3><p>John concludes that this miracle “revealed his glory.”</p><p>The miracle wasn’t merely about solving a social problem.</p><p>It revealed identity.</p><blockquote>Likewise, the highest purpose of our gifts is not self-promotion.</blockquote><blockquote>It is to reveal something greater than ourselves.</blockquote><p>Achievement without revelation is merely success.</p><blockquote>Achievement that reveals purpose becomes legacy.</blockquote><h3>6. Final Reflection</h3><p>The miracle at Cana teaches that extraordinary lives are built through ordinary faithfulness.</p><blockquote>Show up.</blockquote><blockquote>Pay attention.</blockquote><blockquote>Think differently.</blockquote><blockquote>Serve quietly.</blockquote><p>Trust that your purpose will reveal itself in its appointed time.</p><p>Not everyone who walks with you today will see tomorrow’s glory.</p><p>But your responsibility is not to control the audience.</p><blockquote>It is simply to become who you were created to be.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/543/1*WMeJ3824hbOi1uNI2sn6fA@2x.jpeg" /></figure><p>About the author: Dr. Michael Akinwumi writes essays at the intersection of sacred texts, math, leadership, philosophy, and responsible innovation. He meditates on inspired sacred literature through mathematics, philosophy, leadership, and systems thinking.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=96a19caed27e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tower of Babel or Pentecost? Magnificent Humanity and the AI Accountability Gap]]></title>
            <link>https://medium.com/@datawumi/tower-of-babel-or-pentecost-magnificent-humanity-and-the-ai-accountability-gap-b5d16be490b3?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/b5d16be490b3</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[governance]]></category>
            <category><![CDATA[ai-safety]]></category>
            <category><![CDATA[geopolitics]]></category>
            <category><![CDATA[policy]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Wed, 27 May 2026 06:55:29 GMT</pubDate>
            <atom:updated>2026-05-28T01:15:23.853Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A man in America asked three AI systems a simple financial question. He got three different answers. That gap — $587.82 wide — is a parable for our moment.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Jv0vTAxzmMc4eAS6_Jp3Nw.png" /></figure><p><strong>A TEST CASE HIDDEN IN PLAIN SIGHT</strong></p><p>He was not asking for philosophy. He was not asking for a prediction about the future of artificial intelligence, or a geopolitical assessment, or a moral framework. He was a business owner in America preparing a profit-and-loss statement for his consulting practice. He fed identical financial data to three different AI systems and asked each one the same question.</p><p>He got three different answers.</p><p>The spread between the highest and lowest estimate was $587.82 — roughly 8.5% of the mean. For a small business owner making decisions about payroll, taxes, or investment, that gap is not a rounding error. It is a meaningful discrepancy with real consequences.</p><blockquote><em>Two models clustered tightly. One sat nearly $600 lower. That is not noise. That is a signal — and it demands an explanation.</em></blockquote><p>He wanted the truth based on the factual financial data he provided. He did not want ideological manipulation, however subtle. He wanted accuracy. What he got was a demonstration of exactly the accountability gap that Pope Leo XIV is warning us about.</p><p>His experience is a parable for our moment. Not because AI made a mistake — systems make mistakes — but because there is currently no robust mechanism to identify who is responsible for that mistake, to challenge it, or to remedy any harm it causes. The opacity of the internal processes leading to a result makes it “harder to assign responsibility and correct errors.”</p><p><strong>THE POPE ENTERS THE CHAT</strong></p><p>On the day after Pentecost Sunday, Pope Leo published his first encyclical. It was titled <a href="https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html">Magnifica Humanitas</a> (or “Magnificent Humanity”), largely a call to prevent AI from dominating or replacing humankind.</p><p>Some will roll their eyes. Others will argue, as some already have, that the Vatican should stick to matters of morality. But here is the thing: that is precisely what this is. The Magnifica Humanitas is a moral document responding to a moral crisis — and the distinction it draws is one that secular policymakers have struggled to articulate.</p><p>The Pope frames the challenge through two biblical images. On one hand, the Tower of Babel: a collective project that follows a plan which dominates and ultimately dehumanizes. On the other hand, the rebuilt walls of Jerusalem — reconstructed piece by piece under Nehemiah’s direction as a project of shared responsibility.</p><blockquote><em>“Humanity is today facing a pivotal choice: either to construct a new Tower of Babel or to build the city in which God and humanity dwell together.”</em></blockquote><p>You do not need to be Catholic — or even religious — to recognize the force of that image. The Tower of Babel is not a story about ambition. It is a story about what happens when power becomes concentrated, when a singular logic dominates, when the diversity of human experience is flattened in service of one overriding goal. Sound familiar?</p><p>The encyclical calls for slower development of AI and robust legal frameworks. It warns of the risks of attaching AI to weapons. It warns of the danger of social control arising from mass collection of personal data. And it calls for AI to be, in the Pope’s striking phrase, disarmed — freed from “the mentality of armed competition that drives the race for ever more powerful algorithms and larger datasets.”</p><p>The pontiff’s reach extends to 1.4 billion baptized Catholics. But his argument reaches further than his congregation. If national governments, world democracies, and corporate leaders are truthful about their intentions to use AI to benefit all humans, they cannot afford to dismiss a moral framework simply because it arrives wearing a cassock.</p><p><strong>WHAT ACCOUNTABILITY ACTUALLY REQUIRES</strong></p><p>The man’s experience about his P&amp;L statement raises an important question: which AI was right? The deeper question is: how would he know? And who is responsible for helping him find out?</p><p>The opacity of AI systems’ internal processes makes it harder to assign responsibility and correct errors. As a result, fundamental scientific aspects such as the internal representations and computational processes of these systems remain, at present, unknown even to their creators.</p><p>Entrusting an algorithm with the power to determine who is creditworthy, who is hireable, who receives a loan, who gets flagged at a border — without anyone bearing clear responsibility for that judgment — is to hand over the task of redefining the boundaries of human possibilities. AI embodies choices and priorities through what it “measures, what it ignores, what it optimizes, and how it classifies people and situations.”</p><blockquote><em>“It is not enough to invoke ethics in the abstract. Robust legal frameworks, independent oversight, informed users, and a political system that does not abdicate its responsibility are required.”</em></blockquote><p>The White House’s own principles for AI in the federal government speak of truth-seeking and ideological neutrality — that LLMs should be truthful, acknowledge uncertainty, and refrain from manipulating responses in favor of ideological agendas. These are good principles. But principles without enforcement mechanisms are aspirations. The man asking three AI systems for his P&amp;L total was, in a sense, conducting an informal test of those principles. Two of the three systems passed. One did not. He has no way to know which is which.</p><p>That is the accountability gap. And closing it — through rights to explanation, rights to contest, clear legal liability, and mandatory pre- and post-deployment auditing — is not a concession to AI doomers or a capitulation to regulation for its own sake. It is the engineering minimum for a technology deployed at scale in a democratic society.</p><p><strong>PENTECOST, NOT BABEL</strong></p><p>The Pentecost story, as a counterpoint to Babel, is about the restoration of communication across differences. Where Babel produces fragmentation through the confusion of tongues, Pentecost produces understanding — not uniformity, but genuine comprehension across diverse peoples. It is, at its core, a story about technology — the technology of language — being used not for domination but for connection.</p><p>AI presents the same fork in the road. The question is not whether the technology is good or bad. It is not good, bad, or neutral. It presents an opportunity for stakeholders — developers, deployers, governments, civil society, and users — to demonstrate responsibility. The question is whether we will govern it as a shared project, built piece by piece with accountability and care, or whether we will race toward a Tower whose height we cannot control and whose consequences we cannot predict.</p><p>The man with his three P&amp;L statements deserves an answer. So do the seven in ten Americans who fear for their jobs. So do the communities that have long been subjected to algorithmic denial and automated harm. So do the workers in every country watching the geopolitical AI race accelerate with no meaningful safety net in place.</p><p>Magnificent humanity — if we choose to be magnificent — can responsibly harvest the benefits of AI. But that word responsibly is doing enormous work. It requires legal frameworks. It requires accountability. It requires the kind of shared responsibility that rebuilds walls piece by piece, rather than the kind of collective ambition that builds towers until they fall.</p><p><strong>The choice is ours. And it is urgent.</strong></p><p>✦ ✦ ✦</p><p><em>Dr. Michael Akinwumi is the Chief AI Officer &amp; Head of the Responsible AI Lab at the National Fair Housing Alliance. He is also an Adjunct Professor of Responsible Machine Learning at George Washington University, School of Business, and the Co-Founder and Chief Mathematical Scientist at Neryx Applied Intelligence. The views expressed in this article are solely his own and do not represent the positions or policies of any of the organizations with which he is affiliated.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b5d16be490b3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Build the Antilock Brakes]]></title>
            <link>https://medium.com/@datawumi/build-the-antilock-brakes-7d97d4732b41?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d97d4732b41</guid>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Sat, 23 May 2026 11:48:35 GMT</pubDate>
            <atom:updated>2026-05-23T11:48:35.726Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9GcRI-mVKB5nMDDdghC8oA.png" /></figure><p><em>The cancelled executive order on AI testing wasn’t a threat to innovation. It was a missed opportunity to lead.</em></p><p>No reasonable government would ground a safe commercial aircraft and replace it with a faster, cheaper plane that skips safety inspections. Yet that is essentially the logic behind cancelling a proposed executive order that would have mandated AI testing at the federal level — all in the name of winning a race.</p><p>The central narrative driving this decision — that America must choose between AI innovation and AI safety — is false. And accepting it as true may cost the country far more than any regulatory delay ever could.</p><p>The geopolitical stakes are real. The United States, China, Africa and the European Union are pursuing sharply divergent governance philosophies in a competition that has decisively shifted from collaboration to rivalry. That much is not in dispute. But consider what our so-called competitors have actually done: The European Union has embraced comprehensive regulation through the EU AI Act. China has implemented state-control approaches with robust guardrails on deepfakes and algorithmic recommendation systems. These are the nations America is racing against and both have enacted meaningful AI governance. America alone has yet to pass comprehensive federal AI legislation.</p><p><strong>Racing to the bottom is not winning.</strong></p><p>At home, the situation is equally troubling. Federal civil rights enforcement infrastructure has been weakened. Regulators are withdrawing from meaningful oversight of emerging AI issues at precisely the moment when that oversight is most needed. When national security frames crowd out safety and civil rights frames in the AI race, the tools for cross-jurisdictional accountability deteriorate and America may find itself racing to the bottom among its contemporaries.</p><p>Winning the AI race requires three pods of protections: a right to explanation, so that individuals understand why an AI system made a decision affecting them; a right to contest that decision; and clear legal liability, so that someone is accountable when things go wrong. These would enable pre- and post-deployment auditing — the kind of proactive safety engineering that we demand from every other industry that touches the public.</p><p>Think of it this way: we do not ask automobile manufacturers to simply promise their cars are safe. We require antilock brakes. We mandate crash testing. We hold manufacturers liable when defects cause harm. Regulators and federal agencies need to think and act like engineers — working with AI developers and deployers to build proactive safety features into systems, rather than relying solely on human intervention after problems materialize.</p><p>Critics of the executive order argued it would hand a victory to so-called AI doomers, chill investment, or slow a sprint against China that carries existential stakes. These are not trivial concerns. But they are also not arguments against safety — they are arguments against a caricature of safety. A well-designed AI executive order does not pit the American people against AI developers and deployers. It creates the conditions under which both can thrive.</p><p>There is also a deeper question that this moment demands we answer honestly: America has wounds that span its two and a half centuries of existence — in its history of discrimination, in its uneven distribution of opportunity, in the communities that have long been subjected to surveillance, algorithmic denial, and automated harm. Will AI widen those wounds, or heal them? That question cannot be answered by the market alone.</p><p>The good news is that the United States already possesses the legal frameworks and the civil rights doctrine needed to win the AI race safely. The Administrative Procedure Act, the Equal Credit Opportunity Act, the Fair Housing Act, constitutional due process protections — these are not relics. They are the infrastructure of a society that decided, over generations, that power must be accountable. Now is the time to apply, enforce, and extend these laws and principles in ways that continue to position America as the leader of a safe world.</p><p>Pitting legitimate concerns about safe AI development against a free-market approach is not a winning strategy. It is a false choice and one that history will not judge kindly.</p><p>The White House has an opportunity to demonstrate that safety and speed are not enemies. Issue the executive order. Require testing. Build the antilock brakes into the system. That is not losing the AI race. That is how you win it and deserve to.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d97d4732b41" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The King’s Pattern]]></title>
            <link>https://medium.com/@datawumi/the-kings-pattern-229403639cfa?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/229403639cfa</guid>
            <category><![CDATA[algorithmic-bias]]></category>
            <category><![CDATA[social-justice]]></category>
            <category><![CDATA[women-in-tech]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ethics]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Sun, 22 Mar 2026 02:22:47 GMT</pubDate>
            <atom:updated>2026-03-22T21:40:49.785Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><strong>In Honor of Women’s History Month — March 2026</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BunEBBfYLcOuRfr_uWXEaQ.jpeg" /><figcaption><em>Nebuchadnezzar, Meritocracy, and the Hidden Bias in Machines</em></figcaption></figure><p>Women’s History Month is, among other things, a record of institutional exclusion. It reminds us of the long history of systems designed by those already inside them to produce more people who look like them. The suffragists excluded from polling places. The women scientists uncredited. The female executives turned away. This essay is about what happens when those selection criteria get encoded into a machine.</p><p>Nearly 2,600 years ago, King Nebuchadnezzar of Babylon enumerated exactly what “best” meant in a candidate: <em>“…young men without any physical defect, handsome, showing aptitude for every kind of learning, well informed, quick to understand, and qualified to serve in the king’s palace.” — Daniel 1:4 (NIV)</em>. If alive today with a data science team, he would have trained a machine learning classifier on those criteria and deployed it at scale. Babylon, optimized.</p><p>But within this definition of meritocracy is a minefield of protected characteristics, proxy variables, and civil rights violations and in the current political moment, as the federal government dismantles the very frameworks designed to catch this kind of prejudice, the stakes could not be higher.</p><h4>A Single Phrase, Three Protected Classes</h4><p>“Young men without any physical defect” is three seeming violations stacked on top of one another. Age (“young”) is protected under the Age Discrimination in Employment Act (ADEA) for workers 40 and older. Sex (“men”) is protected under Title VII of the Civil Rights Act and it is the single word that would have filtered out Harriet Tubman, Shirley Chisholm, and every woman Women’s History Month exists to honor. Disability (“without any physical defect”) is protected under the Americans with Disabilities Act (ADA). A model trained on this phrase does not learn three independent penalties; it learns their intersection, thereby compounding disadvantage for anyone who sits at the crossing of all three, in exactly the way Kimberlé Crenshaw’s intersectionality framework predicts.</p><p>“Handsome” compounds the problem. Appearance ratings are proxies for race, sex, disability, and age. Removing a protected characteristic from the feature set does not remove the discrimination when a proxy remains in the model. It merely obscures it. “Well informed” encodes the knowledge base of whoever defines what counts as knowledge, making it a potential proxy for national origin and class.</p><p>As a society we need to be woken up to the same mechanism that kept women out of the professions for generations: defining qualification as resemblance to those already qualified.</p><h4>The King’s Pattern Meets Executive Orders</h4><p>Nebuchadnezzar invoked meritocracy. His criteria were framed as objective standards of excellence. He was not, in his own telling, discriminating against women, people with disabilities, or the aged. He was simply selecting the best. The current administration’s January 2025 executive orders — “Ending Radical and Wasteful Government DEI Programs” (EO 14151) and “Ending Illegal Discrimination and Restoring Merit-Based Opportunity” (EO 14173) — frame DEI in precisely the same oppositional terms: as an obstacle to merit rather than a mechanism for ensuring that merit assessments are not themselves corrupted by bias.</p><p>EO 14173 is effectively taking us back to the pre-1965 era before the Executive Order 11246, the Johnson-era directive requiring federal contractors to affirmatively address barriers to employment. A subsequent order, “Restoring Equality of Opportunity and Meritocracy,” directed federal agencies to cease pursuing disparate-impact claims. This is the decisive move. Disparate-impact liability is precisely the legal doctrine that would catch the king’s pattern. It requires no proof of intent; it asks only whether a neutral-seeming system produces disparity in outcomes and if reasonable efforts were taken to mitigate the disparity. Eliminating its enforcement does not eliminate allocative bias, it only makes it legally invisible. While there are active legal fights against elements and proxies of these orders, the policy signal is clear: institutions that used DEI frameworks to audit their own allocative systems for bias are now under pressure to stop doing the work that would catch the king’s pattern before deployment.</p><h4>Transparency Is a Civil Rights Requirement</h4><p>Transparency rests on three inseparable conditions: the ability to understand how a system decides, the ability to explain why it decided as it did, and the ability to contest what it decided. Interpretability. Explainability. Contestability. Deny any one and you have denied accountability itself. In a regulatory environment where enforcement is being narrowed, disparate-impact doctrine is being gutted, and DEI audit infrastructure is being systematically dismantled, these three conditions are not a backstop. They are the last available ground on which accountability can stand. When the king’s pattern cannot be interrogated, the king decides everything.</p><p>Tools like SHAP can attribute prediction scores to individual features, giving a rejected applicant some basis for challenge. But SHAP assumes feature independence, which fails exactly where intersectionality matters most — where age, sex, and disability interact. An explanation that says “your age reduced your score” while hiding that the age penalty was amplified by sex and disability is not merely incomplete. It is misleading. A model that can explain itself can be interrogated. A model that can be interrogated can be audited. A model that can be audited can be challenged. In a society governed by civil rights laws, the right to challenge a decision is not a technical feature. It is a constitutional one.</p><h4>The Women the King Did Not Select</h4><p>Women’s History Month is not a celebration. It is a reckoning. Rosalind Franklin did not lack aptitude for every kind of learning. Katherine Johnson was not slow to understand. Shirley Chisholm was not unqualified to serve. They did not fail the king’s pattern. The king’s pattern recognition system failed them. And every woman whose name this month exists to honor was filtered out not by a neutral standard but by a king’s pattern — a system that mistook its own blind spots for objective criteria, encoded its own limitations as universal truth, and called the result meritocracy. That is not ancient history. That is the architecture of every unchecked algorithm running today.</p><p>Here is what I want you to remember. When we dismantle disparate-impact enforcement, we remove the only legal doctrine that says it is enough to begin our quest to equity by showing the machine is producing outcomes a just society cannot accept. When we shutter DEI audit infrastructure, we cut the institutional memory of who the system has been leaving out, and why. When we call all of this a restoration of meritocracy, we are telling the oldest story in the archive: a standard designed to select people who look like the people who designed the standard, now automated, scaled, and legally insulated from scrutiny at the same moment.</p><p>Allowing allocative bias to thrive has a compounding effect. It gets to a critical point where the system can no longer imagine excellence in a form it has not already approved after decades of seeing each generation of the king’s pattern trained on the previous generation’s exclusions. The best allocative system is the one wise enough to interrogate its own criteria before it runs. The best society is the one that insists on that interrogation even when — especially when — it is politically pressured not to.</p><p>This Women’s History Month, do not merely remember the women the king did not select. Demand the audit. Require an explanation. Stand up for something. The king is still sorting. The question is whether we are finally wise enough to check his work.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=229403639cfa" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Exclusion from Mathematics Is Exclusion from Power: A Semiquincentennial Reckoning]]></title>
            <link>https://medium.com/@datawumi/exclusion-from-mathematics-is-exclusion-from-power-a-semiquincentennial-reckoning-327f3cf7ebee?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/327f3cf7ebee</guid>
            <category><![CDATA[history]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[black-history-month]]></category>
            <category><![CDATA[mathematics]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Mon, 16 Feb 2026 03:01:14 GMT</pubDate>
            <atom:updated>2026-02-16T03:01:14.785Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ick2zQP3Nu0-Snh4ull0WA.jpeg" /></figure><p>In 1926, historian Carter G. Woodson founded what was then called Negro History Week and later evolved to the Black History Month. His intention was both simple and radical: to insist that Black achievement be studied, documented, and remembered in a nation that too often refused to record it. A century later, as mathematics quietly governs the architecture of artificial intelligence, finance, logistics, and public policy, the question before us is not merely historical.</p><p>It is structural.</p><p>What does it mean to celebrate history in a field that rarely records or amplifies your presence?</p><p>Mathematics is not ornamental to modern society. It is foundational. It underpins machine learning models, optimization systems, probability engines, and predictive tools that now shape access to housing, credit, employment, and opportunity. If representational harm exists in AI systems, it is inseparable from the mathematical abstractions beneath them. Addressing this harm requires investing in the representativeness of the professionals that formulate the foundations of the disciplines from which these systems emerge.</p><p>To commemorate Black excellence in mathematics, then, is not symbolic. It is strategic, because exclusion from mathematics is exclusion from structural power.</p><p>This is more important as we look forward to the U.S. Semiquincentennial.</p><h3>The Era of Firsts: Visibility Against Resistance</h3><p>In the summer of 1925, <strong>Elbert Frank Cox</strong> completed his dissertation on polynomial solutions of the difference equation<br><em>af(x+1) + bf(x) = φ(x)</em>.</p><p>He became the first Black American to earn a Ph.D. in mathematics and likely the first Black man in the world to do so. That achievement was structural defiance. In an era when access, not intellect, was restricted, Cox demonstrated that excellence was never the constraint.</p><p>Soon after, <strong>Euphemia Lofton Haynes</strong> became the first Black American woman to earn a Ph.D. in mathematics. Her dissertation explored independent conditions characterizing symmetric correspondences. Beyond theory, she reformed educational systems in Washington, D.C., linking abstraction to institutional change.</p><p><strong>William Schieffelin Clayton</strong> broke barriers in topology and became the first Black American to publish in the prestigious <em>Annals of Mathematics</em>. <strong>Mary Winston Jackson</strong> advanced aeronautical engineering at NASA, contributing to early space research while advocating for broader inclusion in STEM.</p><p><strong>David Blackwell</strong> reshaped probability, decision science, and game theory, fields that now anchor economic design and AI decision systems. In 1947, he introduced what is now known as the Rao–Blackwell Theorem, a foundational result showing how statistical estimators can be systematically improved using conditional expectation. He architected a deep connection between probability, game theory, and rational choice under uncertainty.</p><p>These were not isolated achievements. They were proofs.</p><p>Each theorem published, each peer-reviewed paper accepted, signaled that intellectual capacity was never racially constrained. Access was.</p><p>Excellence survived in systems not designed to nurture it.</p><h3>Mathematics as Measurement and Revelation</h3><p>Mathematics measures. It models. It predicts.</p><p>But it also reveals.</p><p>Over the last hundred years, the trajectory of Black mathematicians reveals a structural truth: when institutions expand and participation widens, systems strengthen.</p><p>From 1926 to today, the work has evolved from proving theorems to proving stability. Mathematical models now govern consequential decisions affecting millions. Probability distributions shape credit scores. Optimization functions allocate resources. Algorithms influence who is visible and who is invisible.</p><p>Abstraction has become infrastructure.</p><p>The stakes are no longer confined to academic journals. They now determine dignity.</p><p>In this era, mathematics has become the backbone of modern economies. Which means that exclusion from mathematics is exclusion from structural power.</p><h3>Participation in the Conversation of Abstraction</h3><p>My goal in reflecting on this history is not merely to honor “firsts.” It is to acknowledge a century-long expansion of who is permitted to participate in the conversation of abstraction itself.</p><p>Abstraction is not neutral. It encodes assumptions. It shapes incentives. It translates values into systems.</p><p>When Black mathematicians entered the discipline, they did more than solve equations. They expanded the moral imagination of the field. They demonstrated that while the purity of theory and the impurity of institutions can coexist for a period, institutions must evolve.</p><p>We can no longer afford environments that doubt, limit resources, or rarely record or amplify the presence of Black mathematicians.</p><p>The past century shows us what happens when barriers fall. Talent compounds. Systems stabilize. Innovation accelerates.</p><p>This is not a zero-sum story. It is a growth function.</p><h3>From Reckoning to Responsibility</h3><p>As we commemorate Black History Month — officially recognized in 1976 during the U.S. Bicentennial — we are reminded that celebration alone is insufficient. Recognition must translate into investment.</p><p>The next generation of AI will be built on mathematics.</p><p>The question is: who will shape its assumptions?</p><p>If mathematics underpins AI, and AI underpins modern society, then the inclusion of diverse mathematicians is not a cultural preference. It is a systemic necessity.</p><p>To celebrate Black excellence in mathematics is to affirm that the future of abstraction must be expansive.</p><h3>Proof and Possibility</h3><p>The century since 1926 offers proof of persistence. It also offers a blueprint.</p><p>The mathematicians who came before did not wait for permission to be excellent. They wrote their presence into journals, equations, and institutions. They shifted priors — not through rhetoric, but through rigor.</p><p>Their legacy is not confined to history. It lives in every model that works better because participation widened.</p><p>In a time when mathematical systems increasingly mediate opportunity, we must dismantle barriers not only for fairness, but for system resilience.</p><p>Excellence survives. But it should not have to survive against resistance.</p><p>It should be cultivated.</p><p>A century after Woodson insisted that Black achievement be recorded , we are called to do more than record. We are called to build.</p><p>Because when participation in abstraction expands, possibility expands with it.</p><p>That is not symbolic.</p><p>It is mathematical.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=327f3cf7ebee" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[October Highlights: Advancing Responsible AI Across Policy, Industry & Global Dialogue]]></title>
            <link>https://medium.com/@datawumi/october-highlights-advancing-responsible-ai-across-policy-industry-global-dialogue-0e0adf39f80b?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/0e0adf39f80b</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[public-policy]]></category>
            <category><![CDATA[governance]]></category>
            <category><![CDATA[society]]></category>
            <category><![CDATA[innovation]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Mon, 03 Nov 2025 20:07:06 GMT</pubDate>
            <atom:updated>2025-11-03T20:54:21.102Z</atom:updated>
            <content:encoded><![CDATA[<p>October was a month of meaningful conversations on how we can shape AI systems that are not only innovative but also accountable and inclusive.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*doeQDbWeBmwMgIaOtZ3WzA@2x.jpeg" /></figure><h3>Panel: “The Good, the Bad, and the Work Ahead”</h3><p>At an invite-only convening, I joined researchers, policymakers, and state AG representatives to explore how states can move from reactive to proactive AI governance.</p><p>Key takeaway: Collaboration between developers, deployers, and regulators builds durable safeguards before harms occur.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*yf-7R5M6iezMKt5KFug7pQ@2x.jpeg" /></figure><h3>Impact Maryland: Building an AI-Ready Workforce</h3><p>At Impact Maryland — hosted by the Baltimore Banner — I engaged in timely discussions on how Maryland is preparing its workforce for an AI-driven economy amid shifts in federal research funding.</p><p>Key takeaway: Resilience and inclusion are the cornerstones of sustainable AI workforce readiness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*WnYu8WC4EWlwbFuawd-yqg@2x.jpeg" /></figure><h3>Industry Visit: Inside the Vision for AI Agents</h3><p>During an on-site visit with a major player in the global AI field, particularly noted for its open-source approach to large language models (LLMs), I listened and contributed to conversations about the company’s long-term vision for AI and the evolving relationship between innovation and governance.</p><p>Key takeaway: Responsible innovation is achievable when transparency and long-term thinking guide development.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*XNBzKT1sMSKf9_BAfUYW6g@2x.jpeg" /></figure><h3>Global Dialogue on AI Governance</h3><p>I concluded the month with a panel at the Consulate General of India in New York, discussing how countries without legacy tech infrastructure can effectively leverage U.S.-origin AI models and infrastructure to foster responsible, inclusive growth.</p><p>Key takeaway: Responsible AI must be a shared global endeavor that elevates emerging economies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rhOYOoZTEOfJr9JR3BED2w@2x.jpeg" /></figure><h3>Collaboration, Competition &amp; Community</h3><p>After more than five years, I finally played ping pong with some of the brightest minds — a reminder that innovation thrives when we make space for connection and joy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ofzqob_JcPh23n2GsaWvtw@2x.jpeg" /></figure><h3>Looking Ahead</h3><p>As we look toward the months ahead, I remain committed to aligning AI innovation with civil and human rights principles — ensuring that fairness, transparency, and accountability remain the foundation of every technological advance.</p><blockquote>Progress in AI is most meaningful when it strengthens both human dignity and public trust.</blockquote><p>Follow for future updates: @datawumi on X and BlueSky</p><p>#ResponsibleAI #AIInnovation #AIGovernance #AIPublicTrust #AIPolicy</p><p>⸻</p><p>Dr. Michael Akinwumi (@datawumi on X and BlueSky) is a math scientist, Chief AI Officer and Head of a Responsible AI Lab. He writes about technology, sociotechnical issues, the intersection of innovation and justice, and human adaptability — exploring how people can navigate change with confidence and clarity.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0e0adf39f80b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Courage to Pivot: Transforming Doubt into Momentum]]></title>
            <link>https://medium.com/@datawumi/the-courage-to-pivot-transforming-doubt-into-momentum-6e07db273124?source=rss-4b97e98bc72c------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e07db273124</guid>
            <category><![CDATA[resilience]]></category>
            <category><![CDATA[personal-growth]]></category>
            <category><![CDATA[change-management]]></category>
            <category><![CDATA[career-change]]></category>
            <category><![CDATA[professional-development]]></category>
            <dc:creator><![CDATA[Michael Akinwumi]]></dc:creator>
            <pubDate>Mon, 27 Oct 2025 18:19:18 GMT</pubDate>
            <atom:updated>2025-10-27T18:19:18.316Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2CAXej1xt4WTS1G8VSAKUg.jpeg" /><figcaption>What path leads to my optimal impact?</figcaption></figure><p>If you’ve ever stared at a job description for a role you genuinely want, only to feel a sinking feeling in your stomach, you know the feeling.</p><p>You read the “Required Qualifications” and a voice in your head starts its relentless monologue: <em>“You’ve only ever done X. They want someone who has done Y. You have no experience. You’re not qualified. You’d be starting from nothing. It’s too late to pivot.”</em></p><p>This is “experience anxiety,” and it’s the single biggest killer of career changes. You look at your current job — managing a retail store, teaching eighth-grade history, processing invoices in accounting — and then look at your desired job — data scientist, project manager, UX designer — and you see an unbridgeable chasm.</p><p>Your approach is wrong. Stop looking at your careers backward.</p><p>One of the oldest stories about courage and reinvention offers a surprisingly modern insight into this struggle. It’s the story of David — once a shepherd, later an elite soldier — who moved from guarding sheep to leading a nation.</p><p>If David’s journey were rewritten today, it might sound like this:</p><p><em>A young professional managing spreadsheets learns to solve complex problems with limited tools. Later, that same problem-solving mindset helps him build a startup, lead a nonprofit, or direct a team. To outsiders, it looks like a leap. In reality, it’s a transfer of </em><strong><em>process mastery </em></strong><em>— from one domain to another.</em></p><p>Career reinvention isn’t a break from your past. It’s a continuation of your preparation.</p><p>Based on my reflections on the story of David, here is how you can overcome your doubt and create that very future you imagine. I hope it helps so many people who have reached out to ask how I moved from where I started into doing what I now do.</p><h4>1. The “Experience” Fallacy</h4><p>When we first meet David on the battlefield, he’s not a soldier; he’s a delivery boy bringing lunch to his older brothers. He sees the challenge: a nine-foot giant who has paralyzed the entire army with fear.</p><p>David, a shepherd, says, “I’ll fight him.”</p><p>The reaction is predictable. His own brother, Eliab, accuses him of being arrogant and negligent. King Saul, the ultimate hiring manager, looks at David and says, “You can’t… you’re only a boy, and I’ve been a warrior from my youth.”</p><p>This is the classic “entry-level job requires 5 years of experience” dilemma. Saul and Eliab are looking at David’s résumé.</p><blockquote><strong><em>Current Role</em></strong><em>: Shepherd.</em></blockquote><blockquote><strong><em>Desired Role</em></strong><em>: Giant-Slayer, Elite Shock Trooper.</em></blockquote><p>The “experience” doesn’t match. From the outside, the skills required for “tending the sheep” have nothing to do with being “part of an elite army.” This is the same disconnect you feel. “I’m a teacher, what do I know about corporate project management?” “I’m in sales, what do I know about coding?”</p><p>David’s response is what every career-changer needs to learn. He doesn’t say, “You’re right, I’m new.” He accurately reframes his entire work history.</p><p>“I’ve been keeping my father’s sheep. When a lion or a bear came and carried off a sheep… I went after it, struck it, and rescued the sheep… I killed both the lion and the bear; this giant will be like one of them…”</p><p>David clearly had tons of experience in the process that keeps a business running. He had experience in risk assessment, courage, precision-striking under pressure, and protecting the vulnerable.</p><p>He just had to convince his new boss that the process for keeping a “lion” out of the business of farming sheep and goats was the same “process” required for keeping a “Goliath” out of national security systems.</p><blockquote>“The process trumps the goals. The patterns in the process should be your goal and not the destination or the starting point.” — Michael Akinwumi</blockquote><h4>2. The Process Trumps the Goal</h4><p>This is the core of the entire argument: <em>The process trumps the goals</em>. We get so hung up on the objects of our work that we forget to value the process. David understood that the objects of his goals were different — a lion on the farm is not the same as a giant on the battlefield. <strong>But he had spent years learning “the process that gets you to the goal, regardless of the environment”</strong>.</p><blockquote>That process — that “system of transferable skills” — is what wins, “irrespective of the target.” — Michael Akinwumi</blockquote><p>Your goal right now should not be a new title. Your goal should be to identify the “patterns in the process” of your current job.</p><p>Stop for a minute and do this. Take out a piece of paper and write down your job title. Now cross it out. Write down the processes you do every single day.</p><blockquote><strong>Are you a barista?</strong></blockquote><blockquote>Wrong description: “I make coffee.”</blockquote><blockquote>Right description (The Process): “I perform rapid-sequence task management in a high-pressure environment. I handle real-time inventory coordination, customer de-escalation, and product quality control simultaneously.”</blockquote><blockquote><strong>Are you a stay-at-home parent?</strong></blockquote><blockquote>Wrong description: “I don’t work.”</blockquote><blockquote>Right description (The Process): “I am the lead coordination, finance, and operations manager for a multi-person household. I handle 24/7 crisis management, nutritional planning, educational curriculum development, and transportation scheduling.”</blockquote><blockquote><strong>Are you an administrative assistant?</strong></blockquote><blockquote>Wrong description: “I answer emails and schedule meetings.”</blockquote><blockquote>Right description (The Process): “I am an executive operations manager. I orchestrate complex, multi-time-zone stakeholder coordination. I serve as the primary information-flow gatekeeper and am responsible for high-level communications and discrete data management.”</blockquote><p>This isn’t just “résumé fluff.” This is the truth. You have been fighting lions and bears for years. You have been mastering a process. The skills are transferable. The only thing that changes is the target.</p><blockquote>Your problem isn’t a lack of skills. Your problem is that you’re describing yourself as a “shepherd” when you’re really a “lion-killer.”</blockquote><h4>3. The Five Stones of Uncertainty</h4><p>Let’s address the fear. It’s one thing to say this. It’s another to walk onto the battlefield.</p><p>When David went to the stream, he “picked five stones.” Why five? He only used one to win.</p><p>We can only speculate. Perhaps he had doubts. Maybe he thought, “What if I miss? What if the first one doesn’t work? What if he’s tougher than the lion?” Perhaps if he knew he only needed one, he would have gone with only one.</p><blockquote>The lesson here is revolutionary: David proceeded even with doubt.</blockquote><p>You do not need to be 100% certain to make a career change. You do not need to have all the answers. This isn’t about blind faith; it’s about courage. <strong><em>Courage is not the absence of fear; it’s the antidote to it. It’s the conviction to act despite uncertainty.</em></strong></p><blockquote>You are allowed to have doubts. You are allowed to pick five stones.</blockquote><blockquote>You just aren’t allowed to let that doubt stop you from walking to the field.</blockquote><p>You will have critics. You just must not be your own critics. David’s own family told him he was unqualified. The king tried to put his own “proven” armor on him, a solution that didn’t fit David’s skills. David was persistent. He managed his own psychology. He projected confidence to those who could destroy it, while internally processing his own uncertainty. This is a critical skill: don’t broadcast your doubts to those who will only amplify them.</p><p>Stop waiting for the fear to go away. It won’t. Pick up your five stones — your doubts, your fears, your “what ifs” — and start walking anyway.</p><h4>4. Get Up from the Dream</h4><p>There is a danger in all this. It’s the danger of “destination dreaming.”</p><p>It’s easy to “sit down or lie down and dream.” We become “preoccupied with sweet descriptions of the promised end.” We spend hours on LinkedIn scrolling through the “perfect” jobs. We fantasize about the bigger salary, the better work-life balance, the corner office.</p><p>But those dreams are inert. “Expect no change until you get up.” Your dream of a new career will not be realized by dreaming. It will be realized by the “strenous rowing and paddling that can get you there.”</p><p>What is the “rowing and paddling”?</p><ol><li>Auditing your “process” like we did in Part 2.</li><li>Learning a new language. Translating “killed a bear” into “managed high-stakes security threats.”</li><li>Practicing your story. Rehearsing, aloud, how your experience as a teacher directly applies to being a scrum master.</li><li>Starting. Applying for the job. Writing the first line of code. Networking with someone in a new field.</li></ol><p>You have to get up from the dream and start rowing.</p><h4>5. Trusting the “Stochastic” Process</h4><p>A “stochastic process” is a term for a path that is governed by a set of probabilities yet appears random and unpredictable. As a math scientist, I consider it a perfect model for a career transition.</p><p>You will have “draws from a stochastic process” that “lead to disappointments.” You will get rejected from your dream job. You’ll fail a certification exam. The company you were targeting will have a hiring freeze.</p><p>These moments feel like contradictions or failures. But in this story, the random path — from pasture to battlefield — was the only path that could have created the right person for the job. The “randomness” was the preparation.</p><p>Every “no” is not a “never.” It’s a “not yet.” Every failed interview is data. Every rejection is a redirection. Randomness is part of your process, not a deviation from it. Your job is to keep mastering your process, trusting that the right “Goliath” will eventually appear.</p><p>And when it does, you must have the mindset to always finish what you start. David didn’t just knock Goliath down; he saw the victory through to its conclusion. When you get your shot, be persistent. Master the new role. Finish what you started.</p><h4>Closing Reflection</h4><p>Every pivot begins with tension — between fear and purpose, between the known and the possible. But the same instincts, disciplines, and values that shaped your past have already prepared you for what’s next.</p><p>The tools in your hand — your skills, habits, and experiences — may look simple now, like a small wooden slingshot. But when you use them with intention, they unlock doors as powerful as any modern key.</p><p>So, take the step. Not because you’re certain, but because you’ve been preparing all along.</p><p>⸻<br><em>Dr. Michael Akinwumi (</em><a href="http://twitter.com/datawumi"><em>@datawumi</em></a><em> on X) is a math scientist, Chief AI Officer and Head of a Responsible AI Lab. He writes about technology, sociotechnical issues, the intersection of innovation and justice, and human adaptability — exploring how people can navigate change with confidence and clarity.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e07db273124" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>