Skip to main content

Why "AI is just the next compiler" doesn't hold up

At a recent talk I compared AI adoption to the introduction of compilers: a new layer of abstraction that lets us work at a higher level, the same way compilers let us stop writing assembly by hand. In the hallway afterwards, a few people pushed back on that comparison. They were right to.

Here's the point I should have made from the stage: agents are not deterministic, and that single difference breaks the analogy.

Compilers follow rules

A compiler is not a magical black box. You give it code, and it gives you a binary, following a fixed set of rules. Given the same input, a compiler will reliably produce the same output, every time. You don't hope the binary does what you wrote.

You trust the compiler, because the transformation is deterministic.

That determinism is exactly what let us move up the abstraction ladder in the first place. We stopped worrying about registers and memory addresses because we could trust the layer below us to behave the same way twice.

Agents aren't deterministic

Ask an agent to implement the same feature five times, and you'll likely get five different solutions. Not five broken ones, five different ones, each with its own trade-offs, its own assumptions, its own blind spots.

That's the part where the compiler analogy doesn’t hold.

Starting to trust AI is not the same as adopting a higher-level programming language.

A language raises the abstraction level while keeping the contract deterministic. An agent raises the abstraction level and removes the contract.

Someone still has to sit on the other side

Agents make coding choices by feeding context and assumptions into a black box of unknown training data. That means someone still needs to sit on the other side of that black box and decide:

  • Which trade-offs make sense for this system, not in the abstract, but for the constraints this specific codebase and team are working under.
  • Whether the implementation fits our technical vision, an agent has no notion of "where we're trying to take this architecture over the next two years."

These tasks require human judgment. It's not something you can compile away.

What this means in practice

I'm not arguing against agentic programming. I use it daily and it's a real productivity gain. But I stopped using "the next compiler" analogy, because that framing implies a level of trust the tooling hasn't earned yet and maybe never will.

The current framing is closer to: agents are a powerful, non-deterministic collaborator. Useful precisely because they generate options a deterministic tool never would and risky for the same reason. The judgment about which of those options is right still belongs to us.

That's it!

Layers of abstraction are only as trustworthy as the guarantees they make.

Compilers made a guarantee.

Agents make a suggestion. (But very good ones)

Popular posts from this blog

Podman– Command execution failed with exit code 125

After updating WSL on one of the developer machines, Podman failed to work. When we took a look through Podman Desktop, we noticed that Podman had stopped running and returned the following error message: Error: Command execution failed with exit code 125 Here are the steps we tried to fix the issue: We started by running podman info to get some extra details on what could be wrong: >podman info OS: windows/amd64 provider: wsl version: 5.3.1 Cannot connect to Podman. Please verify your connection to the Linux system using `podman system connection list`, or try `podman machine init` and `podman machine start` to manage a new Linux VM Error: unable to connect to Podman socket: failed to connect: dial tcp 127.0.0.1:2655: connectex: No connection could be made because the target machine actively refused it. That makes sense as the podman VM was not running. Let’s check the VM: >podman machine list NAME         ...

Cache stampede: when our cache turned against us

While investigating some performance issues, we ran into an ASP.NET Core API that cached a fairly expensive aggregation query for 60 seconds. Under normal load, that was fine: one request rebuilds the cache, everyone else reads from it. Under peak load, dozens of requests would arrive in that same expiry window, all see a cache miss, and all fire the same expensive query in parallel. The database didn't like that. That was the moment when our caching layer stopped helping and started hurting. A burst of requests comes in at the same time, all miss the cache, and all go hammer the database or the downstream API at once. That's a cache stampede . The cache was supposed to protect our backend, and for a few hundred milliseconds it did the opposite. Why this happens IMemoryCache.GetOrCreate (and its async sibling) looks like it protects you, but it doesn't add any locking on its own. Look at the naive version: public async Task<Report> GetReportAsync(string key) ...

Azure DevOps/ GitHub emoji

I’m really bad at remembering emoji’s. So here is cheat sheet with all emoji’s that can be used in tools that support the github emoji markdown markup: All credits go to rcaviers who created this list.