Is Learning to Code Still Worth It in 2026?

If you are _already_ a [software engineer](https://en.wikipedia.org/wiki/Software_engineering) or _aspiring_ to be one,
watch this video by Alberta Tech: https://youtu.be/42V0xazKHZA
It should reassure you - even just temporarily - that there is still a "future" for coding
and it's a worthwhile investment of your time.

She makes several valid points.
1. Ai Model CEOs (with a vested interest in hyping) have be predicting the end of software engineering for a _long_ time.

https://x.com/fchollet/status/1993341471039406405

In case you don't know, @fchollet creator of [Keras](https://github.com/keras-team/keras) and [ARC-AGI](https://github.com/fchollet/ARC-AGI) is not a pundit; he's the real deal. 🧠

The Ai Company CEO's/Investors have a vested interest in making highly optimistic predictions to drive their valuations.

2. LLM Creators are running out of primary data (code) to train on.

Yes, the Large Language Models (LLMs) `continue` to improve through refinement and feedback [(RLHF)](https://en.wikipedia.org/wiki/Reinforcement_learning_from_human_feedback)
but there are major [diminishing returns](https://en.wikipedia.org/wiki/Diminishing_returns) which mean the _rate_ of improvement is slowing. 📉
At this point, _all_ the coding LLMs have _already_ scraped ([stolen?](https://www.theregister.com/2023/05/12/github_microsoft_openai_copilot/) 🤔) and assimilated _all_ of `GitHub`, `GitLab`, `StackOverflow` and the blogs/tutorials/videos on the internet. What `else` is left for them to be trained on? 🤷‍♂️

Many of the smartest people in the world and ***Trillions*** of Dollars have been poured into replacing software engineers.

> For good insight into the AI Capex Boom, read:
https://www.bridgewater.com/research-and-insights/the-macro-implications-of-the-ai-capex-boom

Salesforce fired **`4000`** senior engineers last year claiming their "Agentforce" was already good enough to do much of the work.
Salesforce recently admit they were ***wrong***: https://youtu.be/TbIPUVkImGE the "Agentforce" replacing engineers isn't going very well.

Many in the investor/tech community are saying that the firing of employees in favor of Ai is/was a stock-pump, it clearly backfired.

> **Note**: I'm not saying that Ai won't _replace_ many routine software engineer roles, it definitely will, _eventually_.
> I'm just no longer convinced that it will _replace **all** engineers_ in the _long run_; especially with all the [***`slop`***](https://en.wikipedia.org/wiki/AI_slop) that will need to be fixed.

3. Experienced / Senior Engineers with deep knowledge/understanding of systems/deployment will be needed for the foreseeable future.

In the video Alberta states the following:
"_To debug, is to learn what is **wrong** with the code. To know what is wrong with the code,
you have to **understand** what the code is doing in the **first place**._"

So if you want [continue] to be employed as an engineer, just focus on being the _best_ at your chosen systems / tech stack.
i.e. not much has changed ...
And you will be _expected_ to baby-sit Ai/Agents that "know" less than you ... 💭