A growing sentiment as AI tools become increasingly sophisticated is that learning SQL & Python (and other programming languages), is becoming a thing of the past. I’d venture to argue, however, that the rise of AI actually empowers more people than just analysts / engineers to learn languages like SQL & Python than ever before, as well as that more people should learn these languages than ever before. Data-driven decision making is heading towards the mythical “self-service” destination faster than ever before, and companies that grasp this concept will capitalize on it dramatically over the next 5+ years.
select *
from learning_sql_and_python
where motivation = 'I care about doing my job as well as I can!'
The AI Paradox: More Power, More Risk
There’s no denying that AI has democratized access to data analysis in remarkable ways. Tools like ChatGPT, Claude, and GitHub Copilot can generate SQL queries, write Python scripts, and even explain complex statistical concepts in plain English. For the first time in history, you don’t need to memorize obscure syntax or spend hours debugging nested subqueries to extract insights from your data. Surprisingly, though, the easier it becomes to generate code, the more critical it becomes to understand what that code is actually doing.
When you blindly accept what AI produces, you’re essentially trusting a black box with your organization’s decision-making. And while these tools are impressive, they’re far from infallible. AI-generated queries might look perfectly reasonable at first glance, but are prone to contain subtle logical errors like joining tables on the wrong keys, aggregating data at the wrong grain, or filtering out critical segments of the dataset. If you can’t read SQL or Python well enough to spot these issues, you won’t catch them until they’ve already influenced a bad decision.
That said, I’m not here to lecture why learning SQL & Python is a defense mechanism against AI errors, but rather to make the argument for why learning SQL & Python is more valuable to people outside of roles that would traditionally be considered “data roles” than ever before.
Photo by Claudio Schwarz on Unsplash
The Uber Approach: Data Literacy as Default
When I worked at Uber, one of the most striking cultural elements was the company’s approach to data literacy. In your first week, regardless of whether you were in engineering, operations, marketing, or finance, you had the opportunity to learn SQL. Not because everyone needed to become a data engineer, but because the organization recognized a fundamental truth: in a data-driven company, everyone makes better decisions when they can interact with data directly.
This wasn’t about creating a company full of analysts. It was about reducing friction in the decision-making process. When a product manager could pull their own usage metrics, or when an operations lead could query driver patterns without waiting for a data team ticket to be resolved “in the next sprint” (yeah, sure it will), decisions happened faster and with more context.
The key word here is context. When you understand how to query data yourself, you develop an intuition for what questions are answerable, what data exists, and what the limitations of that data might be. You learn to think critically about sample sizes, to question outliers, and to recognize when your analysis might be missing something important.
Beyond the Data Silo: Building Data-Literate Organizations
Too many organizations still treat their data team as a service department; a place where other teams submit requests and wait for answers. This model creates bottlenecks, delays decision-making, and often results in misaligned analyses because the data team doesn’t have full context on the business problem.
The alternative is to build data literacy throughout the organization. This doesn’t mean everyone becomes an expert analyst, but it does mean establishing a baseline competency that allows people to:
- Ask better questions: When you understand the structure of your data, you can frame more precise questions that are actually answerable.
- Recognize their limitations: Perhaps more importantly, data-literate employees know when they’re approaching the edge of their competency and need to pull in someone with deeper expertise.
- Communicate more effectively: When everyone speaks a common language around data, conversations between business stakeholders and technical teams become dramatically more productive.
- Move faster: For straightforward analyses, people can self-serve without waiting for the data team’s availability.
This is the model we advocate for at South Shore Analytics when we work with healthcare organizations, PE-backed companies and early-stage startups. The most successful implementations aren’t about building the fanciest data architecture, they’re about building systems that empower the right people to make data-informed decisions at the right time.
Photo by Rubaitul Azad on Unsplash
Practical Data Literacy in the AI Age
So what does data literacy actually look like when AI can write your code for you? It’s a shift from writing to reading, validating, and directing.
Reading: You need to be able to read SQL and Python well enough to understand what a query or script is doing. When Claude generates a query for you, can you trace through the logic? Do you understand why it’s using a LEFT JOIN instead of an INNER JOIN? Can you identify which fields are being aggregated and at what level?
The key insight, however, is that you no longer have to memorize syntax to be an efficient user of SQL or Python. It is much more important that you understand what is possible as well as learning about the patterns of common errors that can be produced with these tools.
Validating: You should have enough knowledge to spot obvious errors or questionable assumptions. Does the date filter make sense? Is the query actually answering the question you asked, or something subtly different? Are there edge cases that aren’t being handled?
Directing: Perhaps most importantly, you need to know enough to guide the AI toward better solutions. When the first attempt isn’t quite right, can you articulate what needs to change? Can you break down a complex problem into smaller pieces that are easier to solve?
The Fundamental Concepts That Matter
Even with AI handling much of the syntax, certain fundamental concepts remain critical:
In SQL:
- Understanding table relationships and joins
- Knowing the difference between aggregation levels
- Recognizing when data needs to be filtered, and in what order
- Understanding window functions and when to use them
- Grasping the implications of NULL values
In Python:
- Understanding data structures (lists, dictionaries, DataFrames)
- Knowing when to loop versus vectorize operations
- Understanding basic statistics and what different metrics actually measure
- Recognizing when data needs to be cleaned or transformed
- Understanding the basics of functions and modularity
These aren’t things you need to memorize perfectly, but they are things you need to understand. When you have this foundation, AI becomes a tool that amplifies your capabilities. Without it, AI becomes a crutch that makes you dependent and vulnerable to errors.
Looking Forward
As AI continues to evolve, the bar for what constitutes “basic data literacy” will likely shift. Perhaps in five years, writing SQL from scratch will be as rare as writing assembly code is today. But the need to understand data, to think critically about analysis, and to validate results will only become more important.
The future isn’t about choosing between learning data skills or using AI tools, it’s about using AI tools effectively because you have the underlying knowledge to do so. It’s about building organizations where data literacy is the default, where asking questions of your data is as natural as checking your email, and where the data team’s role shifts from being a bottleneck to being force multipliers for everyone else’s data work.
So yes, learn SQL. Learn Python. Not because you’ll always write everything from scratch, but because understanding these tools is what separates someone who can work with AI from someone who’s merely dependent on AI. In an age where data increasingly drives every business decision, that difference matters more than ever.
Our Vision for Data Education
If you couldn’t have guessed by now, our team at South Shore Analytics places a lot of value on SQL literacy for our employees. Furthermore, we believe in advocating for tools that democratize and simplify these concepts — rather than those that obfuscate or them. For this reason, I wanted to take a moment to drop a quick teaser — we’re building a SQL learning platform at South Shore.
Our approach centers on daily SQL challenges that progressively build your ability to read, understand, and ultimately write queries. The goal isn’t to turn everyone into a database engineer. It’s to build the foundational literacy that makes AI tools truly powerful rather than potentially dangerous.
A sneak preview of what our SQL learning platform is going to look like.
We’re starting with SQL because it remains the lingua franca of data work, but our vision is broader: a comprehensive data literacy platform that helps people understand not just syntax, but statistical concepts, data modeling principles, data visualization best practices and the critical thinking skills that separate good analysis from misleading analysis.
Interested in being one of the beta testers? Shoot us a note at info@southshore.LLC
At South Shore Analytics, we’re building tools to make data literacy accessible and practical for modern organizations. If you’re interested in learning more about our SQL challenge platform or our approach to data stack modernization, please reach out. We’d love to chat about how to make your organization more data-literate.