Blog – Alaka Halder

We are training ourselves to mistake algorithmic outputs for moral wisdom

On 24th September 2025, I’m attending the Adam Smith Lecture at Panmure House, where Brendan McCord will be speaking on The Artificial Impartial Spectator: Adam Smith, AI, and the Corruption of Moral Sentiments. I’m grateful to be sponsored by the Cosmos Institute, which selected two attendees through an essay contest. My essay is below.

Adam Smith said “uniformity of [a worker’s] stationary life naturally corrupts the courage of his mind.” How does this risk manifest itself in modern life?

We are training ourselves to mistake algorithmic outputs for moral wisdom, corrupting the very faculty Smith considered essential for human flourishing.

Smith’s factory worker cycled through small, repetitive tasks with little opportunity to develop mental courage. Today’s corruption is more subtle. We feed symptoms to AI, receive polished playbooks, and convince ourselves that we’ve deliberated after making minor adjustments. We sidestep the difficult work of moral reasoning while maintaining its appearance.

Courage requires wrestling with a problem from start to end. A mayor confronting poverty and difficult budgeting decisions must identify the underlying problems. Then she must ask: Whose needs should be prioritized? What kind of community are we building? How do we balance today’s needs against tomorrow’s? To solve these problems well, she needs a deep understanding of her community, a forward-looking moral vision, and ownership of outcomes.

With AI, she can enter demographic data and budgets, adjust the results slightly, and present them as her own. She avoids deeply examining the underlying values that should guide such decisions. If the plans fail, she can shirk responsibility. But she will also lack the practice, knowledge, and courage to tackle the new, unfamiliar set of problems before her.

This corruption is compounded by mistaking AI’s consistency for Smith’s “impartial spectator,” the cultivated moral sense that guides proper judgment. The impartial spectator emerges through self-knowledge and hard-won moral experience. AI offers opaque aggregations whose reasoning and context we cannot easily interrogate. Algorithmic steadiness is not conscience.

A society fluent in selecting and editing pre-packaged ethical decisions will lack the courage to go against consensus and stake out new moral ground when it matters. By retaining the appearance of moral deliberation while abandoning its practice, it will soon discover it has neither the courage to face genuine crises nor principles worth defending.

What do you enjoy about working in industry?

My family is full of academics and can be skeptical of research done in industry, as if its more pragmatic nature makes it less pure or interesting.

Growing up in that environment, I was insulated from the reality that many real-world problems are tedious, have solid but not especially novel solutions, and are still important, interesting, and satisfying to tackle. I first realized this while wrangling data in fintech and benefits administration.

What do you enjoy about working in industry? A family member asked me *the* question again recently.

Breaking into Healthcare Data Roles

𝗛𝗼𝘄 𝗱𝗼 𝗜 𝗯𝗿𝗲𝗮𝗸 𝗶𝗻𝘁𝗼 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲 𝗱𝗮𝘁𝗮 𝗿𝗼𝗹𝗲𝘀 𝘄𝗶𝘁𝗵 𝗮 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝘁 𝗯𝗮𝗰𝗸𝗴𝗿𝗼𝘂𝗻𝗱? Soon-to-be graduates and early-career professionals have asked me this recently. Some thoughts:

𝗗𝗲𝗰𝗶𝗱𝗲 𝘄𝗵𝗮𝘁 𝗮𝘀𝗽𝗲𝗰𝘁𝘀 𝗼𝗳 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲 𝘆𝗼𝘂 𝗰𝗮𝗿𝗲 𝗮𝗯𝗼𝘂𝘁. If you’re drawn to the mission (say, helping patients find high-quality doctors) but don’t care to work hands-on with clinical data, there are analytics, product, and engineering roles that don’t require a healthcare or clinical background. I’ve seen people move into these roles at every level, including senior leadership, as long as they have strong technical and transferable skills.

𝗖𝗲𝗿𝘁𝗮𝗶𝗻 𝗿𝗼𝗹𝗲𝘀 𝗿𝗲𝗾𝘂𝗶𝗿𝗲 𝘀𝗽𝗲𝗰𝗶𝗮𝗹𝗶𝘇𝗲𝗱 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲. Some data science, machine learning, and clinical research roles require a PhD or deep professional experience. If you’re unsure from job descriptions, you can reach out to employees in similar roles. If you have a career path in mind, find someone who started from where you are now and ask about their journey. A short email with 1–3 clear, targeted questions will get a quicker reply than a vague request for a call.

It’s OK to Drop Low-Impact Research Projects

It’s OK to cull seemingly good ideas with low impact. I’m not talking about politics, just research projects at work. I’m lucky to work somewhere that actively encourages this.

Yes, it’s disappointing to discover that an algorithm or product isn’t worth building. I’ve seen this especially demoralize early-career teammates who feel their career advancement depends on publishing or completing a discovery project that leads to a build.

But it’s a success to drop the doomed project early! You completed your goal (research) and reached a clear conclusion. You saved time, money, and effort. Your team is now free to focus on something that matters (or chase down the next shiny new thing!).

I’ve never seen this kind of “failed” discovery work celebrated or highlighted as a “win” on a resume.

Have you? If not, maybe it’s time to start.

Resume and Outreach Advice for Early-Career Data Roles at Startups

I’m excited to see so much interest in the Garner Health data science roles I recently shared on LinkedIn, especially from early-career professionals. Many of you have asked about how to stand out when applying to startups – whether it’s your first job, you’re moving from a large company, or you’re switching industries.

I wish I had time to respond to every message. Instead, I’m sharing resume and outreach advice for early-career data roles, drawing on 7+ years of experience as a hiring manager and interviewer. These views are my own and don’t represent those of past or present employers.

Real humans are reading your resume
At startups, someone on the hiring team will actually read your resume. There’s no need to cram it with keywords intended to bypass automated filters, which are more common at larger companies. Instead, ask yourself or a trusted friend, “If someone read my resume for 10-20 seconds, would they see why I’m a good fit for this role?”

Only list relevant skills
It’s tempting but unnecessary to include every tool you’ve touched. Tech stacks vary widely but you’ll be able to onboard as long as you’re familiar with the core languages and best practices. If you know SQL, you’ll adapt to the minor syntax differences between Snowflake and Postgres.

Building a Better Society through Writing and Community Management – With Glen Weyl

When thinking about creating a new community, reflect on the unique intersection of existing communities you represent and who else might be in that same intersection.
– Glen Weyl

This post originally appeared on the Interintellect community blog.

Why do some communities and movements improve society in the long run while others flatline? What can a new, global, and multidisciplinary community like Interintellect learn from movements past and present as we work to improve the quality and impact of the world’s conversations? As a “community of individualists,” we also recognize the importance of diverse, individual contributions and are especially interested in learning how to thrive and cooperate in the resulting world.

To explore these and related topics, we invited Glen Weyl, a name well-known to those working to rethink and rebuild social, political, and technological institutions. One of the world’s leading economists and sociotechnologists, Weyl co-invented groundbreaking systems such as quadratic voting to improve collective decision-making and help societies solve inequality, stagnation, and political instability. Such concerns and possible solutions are explored in his book Radical Markets: Uprooting Capitalism and Democracy for a Just Society, published in 2018.